DEV Community: Khali Sollis

The Attention Trap: When Good Software Design Works Against the People Using It

Khali Sollis — Wed, 14 Oct 2026 19:29:08 +0000

What behavioral science reveals about engagement loops, digital habits, and the responsibility of developers.

The software that doesn't want you to leave

You open an app to check one thing: whether a package shipped, what time the game starts, a reply from a friend. You find it in seconds. Then a post below it catches your eye, and another below that, and when you next look at the clock twenty minutes have gone somewhere you never decided to send them.

Nothing malfunctioned. The feed did precisely what it was built, tuned and measured to do, and those twenty minutes were logged somewhere as a small engineering success.

Software can be excellent at meeting its engagement objectives while producing experiences that conflict with what its users wanted from their time. What happens when software succeeds at keeping us engaged but fails to respect our original intentions?

Engagement in itself is not the problem. Reading a long essay, learning a language or a video call with family are engagement too. The useful distinction is between engagement a person would endorse on reflection and engagement they drift into and later regret.

Public conversation tends to land on the individual. In a July 2026 segment of NPR's TED Radio Hour, journalist Nayeema Raza proposed that three pre-internet habits could help people slow down and break the screen cycle. Personal habits matter, but every habit plays out in an environment, and on a screen that environment is written by developers. This essay looks at that side of the glass.

The architecture of engagement

Almost every feature associated with compulsive use began as a fix for a real problem. Infinite scrolling removed the friction of pagination on small touchscreens. Autoplay spares someone watching a series from reaching for the remote. Notifications tell you a person needs you. Recommenders find the few relevant items among millions. Streaks help people keep up practice they chose.

What these features share is that they remove a decision point. A paginated list ends, and the end is a natural moment to ask whether you want more. Each convenience shortens the distance between continuing and never having decided at all.

Reviewing 43 papers, Monge Roffarello, Lukoff and De Russis catalogued eleven types of what they call attention-capture damaging patterns, infinite scroll and autoplay among them. They distinguish deceptive designs, which mislead, from seductive ones, which deceive no one and simply make continuing the path of least resistance. Most engagement features are of the second kind, which is why intent is a poor guide to their effects.

In a survey of 120 YouTube users, Lukoff and colleagues found that autoplay and recommendations primarily undermined people's sense of agency, while search and playlists supported it. That is self-reported perception, not measured harm, but it shows that one product can contain features that hand control to the user and features that quietly take it back.

Underneath the interface sits the objective function. A ranking model trained on watch time or click-through rate learns to serve whatever produces those signals. Kleinberg, Mullainathan and Raghavan formalized the difficulty in a theoretical model in which users have an impulsive in-the-moment self and a long-term self with different preferences, while the platform observes only behavior. In their model, raising engagement improves welfare along some directions of design change and lowers it along others, and engagement data alone cannot reveal which. It is a model rather than an empirical finding, but it names the core problem: behavioral data measures what people do, not what they wanted.

What behavioral science actually tells us

The science here is more interesting and less alarming than the popular version.

Habits. Wood and Rünger's review describes habits as responses that, through repetition in stable contexts, come to be triggered by cues rather than deliberate goals. They are an efficient default mode of action, with a curious side effect: people tend to infer from how often they do something that they must have intended it. Software that sits one tap away in every idle moment is well positioned to become such a cue.

Uncertain rewards. Since Skinner's operant conditioning research, psychologists have known that behavior rewarded intermittently tends to persist longer than behavior rewarded every time. A feed mixes things you care about with things you do not, in an unpredictable order, which is why HCI researchers, Rixen and colleagues among them, have discussed partial reinforcement as a plausible explanation for scrolling's pull. It is a plausible hypothesis built on an analogy to laboratory findings, not an established mechanism of feed use.

Dopamine and the word "reward." In 1997, Schultz, Dayan and Montague showed that the activity of dopamine neurons in monkeys matched a reward prediction error, the quantity at the heart of temporal-difference learning algorithms. That paper bridged machine learning and neuroscience, and it has been stretched far beyond what it shows. The signal it describes concerns learning, the updating of expectations, rather than pleasure, and nothing in those recordings establishes that every notification releases dopamine in your brain.

The shared vocabulary causes its own confusion. In a recommender trained with reinforcement learning, "reward" is a number an engineer chose, such as a click or seconds watched, and "learning" means updating parameters. In psychology and neuroscience, the same words describe how organisms' behavior and neural activity change with experience. The mathematics overlaps; the mechanisms differ. What genuinely connects them is the behavioral loop: the system learns what you respond to while you learn what the system delivers. Two learners, coupled through an interface.

Attention. The experimental evidence on interruptions is clearer. In a laboratory study, Stothart, Mitchum and Yehnert found that merely receiving a call or text notification, without answering it, impaired performance on an attention-demanding task. In a randomized field experiment with 237 smartphone users, Fitz and colleagues found that people whose notifications were batched three times a day felt more attentive, productive and in control, and reported better mood, than controls. Hourly batching did little, and no notifications at all brought more anxiety and fear of missing out. The lesson concerns timing and dose, not abolition.

Self-regulation. In an experience-sampling study of 205 adults, Hofmann, Vohs and Baumeister found that desires for media use, along with work, brought about the most self-control failure. A randomized experiment by Allcott, Gentzkow and Song offers stronger evidence. Temporarily paying people to use social media less reduced their use even after the payments stopped, consistent with habit formation, and letting people set limits on their own future use reduced it substantially, suggesting they wanted to use less than they did. Through their economic model, the authors estimate that self-control problems account for 31 percent of social media use: a model-based estimate, but resting on an experimental pattern that is hard to dismiss.

The trouble with "addiction." None of this licenses calling ordinary use an addiction. Reviewing the literature, Panova and Carbonell found insufficient support for smartphone addiction as a disorder and proposed "problematic use" instead. Kardefelt-Winther and colleagues warned that the behavioral-addiction label is being stretched until it pathologizes ordinary leisure.

"Screen time" deserves similar caution. Across three large datasets, Orben and Przybylski found that digital technology use explained at most 0.4 percent of the variation in adolescent well-being; Twenge, Haidt and colleagues contested that reading, and the debate remains open. Meanwhile, a meta-analysis by Parry and colleagues found that self-reported media use correlates only moderately with logged use, and problematic-use scales even more weakly. Total minutes is a blunt instrument. What people do, why, and whether it crowds out the rest of their lives tell us far more.

Why willpower is only part of the equation

If behavior were purely a matter of resolve, interface design would be irrelevant. The Fitz study is a quiet rebuttal: participants did not become more disciplined; their environment changed, and their experience changed with it.

A useful way to see this is to count actions. In a continuous feed, continuing costs nothing, while stopping requires a decision nothing in the interface prompts. In a batched design, continuing costs one deliberate action and stopping costs none. Neither forces anything; they differ in which choice is the default.

A field study by Rixen and colleagues shows how rarely the app itself provides the exit. Tracking 46 participants' social media sessions for a week, they found that most reasons for ending a session had nothing to do with the app: a waiting task, sleep, food, another person. Sessions spent mainly scrolling were rated less positively than those spent mainly posting or watching stories, though individual differences mattered far more than session content.

It is tempting to conclude that friction is the answer. The evidence is more modest. A large meta-analysis of choice-architecture interventions by Mertens and colleagues, subsequently corrected by its authors, reported an average effect, but a reanalysis by Maier and colleagues found no evidence of an overall effect after correcting for publication bias, a disagreement still unresolved. In HCI, Monge Roffarello and De Russis's meta-analysis of digital self-control tools reported a small-to-medium effect on reducing time spent, but noted that the evidence comes mostly from short experiments, often without control groups, that cannot show whether the change lasts. Friction is a design variable worth testing, not a cure.

Two ways to build a content feed

Imagine two versions of the same feed, showing the same posts in the same order. The content is identical. Only the shape of the experience differs.

In Version A, continuous scrolling, new posts appear automatically as the reader nears the bottom of the page. There is no page two, no final post, no visible edge. Reaching the end of what has loaded is never an event, because more has already arrived. The feed never asks whether you want to continue; continuing is simply what happens next. If the reader stops, the decision has to come entirely from them, made against the momentum of the interface.

In Version B, intentional continuation, the feed shows a limited batch of posts, perhaps ten, followed by a visible stopping point: a short note that the batch is finished, a count of how many posts have been seen so far, and an option to load more. Nothing prevents the reader from continuing. The next batch is a single tap away. But the end of the batch becomes a moment, and continuing becomes a choice rather than a default.

That is the essential difference. Version B does not force anyone to stop, impose a time limit or hide content. It restores the kind of decision point that a paginated list or the last page of a magazine once supplied without anyone noticing. In the terms of the previous section, it changes which action costs effort: in Version A, stopping requires initiative; in Version B, continuing does, if only slightly.

Both designs offer access to exactly the same information. What changes is the behavioral environment surrounding it, and the research reviewed above suggests that cues, defaults and stopping points shape behavior in ways that willpower alone does not explain.

This comparison is an illustration, not evidence. Nothing here shows that Version B improves well-being, reduces regret or changes long-term habits. Those are empirical questions, best answered by testing designs like these with real users and measuring the outcomes that matter to them.

Can we measure success differently?

Teams optimize what they measure. Time spent, session length and daily active users are cheap to collect, hard to dispute and tied to revenue. As the Kleinberg model makes explicit, they cannot distinguish a person who stayed because the product was valuable from one who stayed because leaving took effort.

Complementary signals get closer. Task completion asks whether someone who arrived to do something did it. Satisfaction and perceived control, gathered in brief in-product surveys, capture whether the time felt worthwhile. Notification usefulness shows in how often notifications are opened deliberately rather than dismissed or disabled. Voluntary return, sessions started without a prompt, separates a pull the user generates from one the system generates. Reported regret asks directly whether a session was time well spent.

None of these is easy. Surveys suffer from low response rates and selection bias, and regret is colored by mood. Users genuinely disagree: one person's relaxing evening of scrolling is another's lost one. And any metric that becomes a target can be gamed, a well-being metric included.

The practical approach is to treat these as guardrails, not replacements. When an experiment raises session length, check task completion and reported control in the same arms. A change that adds time while lowering perceived control is a different result from one that raises both, even if the headline dashboard cannot tell them apart.

The responsibility of developers

It would be convenient to blame engineers, and wrong. Developers usually implement requirements set elsewhere, under deadlines, against targets and business models they did not choose. Most infinite feeds were built by people trying to make a product feel smooth.

But engineering responsibility has never been limited to intent. We expect developers to consider security risks nobody asked about and accessibility needs missing from the ticket. An interface's behavioral consequences belong in the same category.

In practice that means asking a slightly different question in design review: not only what a feature makes possible, but what it makes easier, more automatic, or harder to interrupt. Engineers can flag metrics that reward behavior users would not endorse. Designers can make stopping points visible. Product managers can propose guardrails. Organizations, which set the incentives, carry the largest share, because individual conscience is a poor substitute for goals that don't penalize doing the right thing.

Designing for the freedom to leave

Engagement and autonomy are not opposites; the best software earns long sessions because people choose them. The problem begins when a system becomes better at keeping people than at serving them, and our metrics cannot tell the difference.

Behavioral science offers no villain here and no simple fix, only a quieter observation: people act within environments, and much of their environment is now made of code. Those who write it are deciding, whether they intend to or not, where the exits are and how easy they are to find.

If we can design software that makes people want to stay, can we also design software that respects their decision to leave?

References and Further Reading

Allcott, H., Gentzkow, M., & Song, L. (2022). Digital addiction. American Economic Review, 112(7), 2424–2463. https://doi.org/10.1257/aer.20210867

Fitz, N., Kushlev, K., Jagannathan, R., Lewis, T., Paliwal, D., & Ariely, D. (2019). Batching smartphone notifications can improve well-being. Computers in Human Behavior, 101, 84–94. https://doi.org/10.1016/j.chb.2019.07.016

Hofmann, W., Vohs, K. D., & Baumeister, R. F. (2012). What people desire, feel conflicted about, and try to resist in everyday life. Psychological Science, 23, 582–588. https://doi.org/10.1177/0956797612437426

Kardefelt-Winther, D., Heeren, A., Schimmenti, A., van Rooij, A., Maurage, P., Carras, M., Edman, J., Blaszczynski, A., Khazaal, Y., & Billieux, J. (2017). How can we conceptualize behavioural addiction without pathologizing common behaviours? Addiction, 112(10), 1709–1715. https://doi.org/10.1111/add.13763

Kleinberg, J., Mullainathan, S., & Raghavan, M. (2024). The challenge of understanding what users want: Inconsistent preferences and engagement optimization. Management Science, 70(9), 6336–6355. https://doi.org/10.1287/mnsc.2022.03683

Lukoff, K., Lyngs, U., Zade, H., Liao, J. V., Choi, J., Fan, K., Munson, S. A., & Hiniker, A. (2021). How the design of YouTube influences user sense of agency. Proceedings of the 2021 CHI Conference on Human Factors in Computing Systems. https://doi.org/10.1145/3411764.3445467

Maier, M., Bartoš, F., Stanley, T. D., Shanks, D. R., Harris, A. J. L., & Wagenmakers, E.-J. (2022). No evidence for nudging after adjusting for publication bias. Proceedings of the National Academy of Sciences, 119(31), e2200300119. https://doi.org/10.1073/pnas.2200300119

Mertens, S., Herberz, M., Hahnel, U. J. J., & Brosch, T. (2022). The effectiveness of nudging: A meta-analysis of choice architecture interventions across behavioral domains. Proceedings of the National Academy of Sciences, 119, e2107346118. https://doi.org/10.1073/pnas.2107346118

Mertens, S., Herberz, M., Hahnel, U. J. J., & Brosch, T. (2022). Correction for Mertens et al., The effectiveness of nudging: A meta-analysis of choice architecture interventions across behavioral domains. Proceedings of the National Academy of Sciences, 119(19), e2204059119. https://doi.org/10.1073/pnas.2204059119

Monge Roffarello, A., & De Russis, L. (2023). Achieving digital wellbeing through digital self-control tools: A systematic review and meta-analysis. ACM Transactions on Computer-Human Interaction, 30(4), Article 53, 1–66. https://doi.org/10.1145/3571810

Monge Roffarello, A., Lukoff, K., & De Russis, L. (2023). Defining and identifying attention capture deceptive designs in digital interfaces. Proceedings of the 2023 CHI Conference on Human Factors in Computing Systems, 1–19. https://doi.org/10.1145/3544548.3580729

NPR. (2026, July 24). Battling screen addiction with three simple habits [Audio segment, TED Radio Hour]. https://www.npr.org/2026/07/24/nx-s1-5887506/battling-screen-addiction-with-three-simple-habits

Orben, A., & Przybylski, A. K. (2019). The association between adolescent well-being and digital technology use. Nature Human Behaviour, 3(2), 173–182. https://doi.org/10.1038/s41562-018-0506-1

Orben, A., & Przybylski, A. K. (2020). Reply to: Underestimating digital media harm. Nature Human Behaviour, 4, 349–351.
https://doi.org/10.1038/s41562-020-0840-y

Panova, T., & Carbonell, X. (2018). Is smartphone addiction really an addiction? Journal of Behavioral Addictions, 7(2), 252–259. https://doi.org/10.1556/2006.7.2018.49

Parry, D. A., Davidson, B. I., Sewall, C. J. R., Fisher, J. T., Mieczkowski, H., & Quintana, D. S. (2021). A systematic review and meta-analysis of discrepancies between logged and self-reported digital media use. Nature Human Behaviour, 5(11), 1535–1547. https://doi.org/10.1038/s41562-021-01117-5

Rixen, J. O., Meinhardt, L.-M., Glöckler, M., Ziegenbein, M.-L., Schlothauer, A., Colley, M., Rukzio, E., & Gugenheimer, J. (2023). The loop and reasons to break it: Investigating infinite scrolling behaviour in social media applications and reasons to stop. Proceedings of the ACM on Human-Computer Interaction, 7(MHCI), Article 228. https://doi.org/10.1145/3604275

Schultz, W., Dayan, P., & Montague, P. R. (1997). A neural substrate of prediction and reward. Science, 275, 1593–1599. https://doi.org/10.1126/science.275.5306.1593

Stothart, C., Mitchum, A., & Yehnert, C. (2015). The attentional cost of receiving a cell phone notification. Journal of Experimental Psychology: Human Perception and Performance, 41(4), 893–897. https://doi.org/10.1037/xhp0000100

Twenge, J. M., Haidt, J., Joiner, T. E., & Campbell, W. K. (2020). Underestimating digital media harm. Nature Human Behaviour, 4, 346–348. https://doi.org/10.1038/s41562-020-0839-4
Wood, W., & Rünger, D. (2016). Psychology of habit. Annual Review of Psychology, 67, 289–314. https://doi.org/10.1146/annurev-psych-122414-033417

When the Reward Disappears: What Neuroscience Can Teach Us About Adaptive AI

Khali Sollis — Tue, 13 Oct 2026 19:02:00 +0000

How a mouse-brain experiment reveals a challenge shared by biological learning and reinforcement learning systems.

The Problem With Yesterday's Winning Move

Picture a simple learning agent with two buttons. Button A pays out 80% of the time; button B pays out 20%. After a few hundred trials, the agent's value estimates settle where you would expect, and it presses A almost every time. By any reasonable metric, it has learned.

Then, without announcement, the payouts swap.

Nothing in the agent's input tells it the world has changed. It still sees two buttons. Its value table still says A is worth about 0.8. So it keeps pressing A, collecting mostly nothing, and its estimate of A drifts downward only as fast as its update rule allows. Depending on how that rule is written, "as fast as it allows" can mean twenty trials or five hundred.

This is the problem hiding inside many deployed decision systems: learning what works is a different task from noticing that what worked has stopped working. A recommender tuned to last year's catalogue or a scheduler calibrated to traffic that a product launch just rewrote may not use reinforcement learning at all, but each carries its training conditions forward as assumptions.

Animals face this problem constantly, and a recent study offers an unusually detailed look at one piece of how a mammalian brain responds when a reliable reward fails to appear.

What the Mouse Experiment Actually Showed

The study, "Spatially heterogeneous acetylcholine dynamics in the striatum promote behavioral flexibility", by Gideon Sarpong, Jeffery Wickens and colleagues, led from the Okinawa Institute of Science and Technology (OIST), appeared in Nature Communications on 17 December 2025. It reached a wider audience about six months later, through an institutional press release carried by ScienceDaily in June 2026.

The task. Head-fixed mice ran on a floating ball through a virtual-reality Y-maze. One arm delivered a drop of sugar water; the other delivered a black screen. Nothing visually marked the rewarded arm, so the mice had to discover it by trial and error. Once a mouse chose correctly in at least 80% of a session's trials, the researchers reversed the contingency—again without any cue. The previously rewarded arm now paid nothing.

Behaviorally, the mice did what you would predict: they kept choosing the old arm for a while (average accuracy after the reversal fell to about 47%), then gradually shifted to the new one.

What was measured. The team expressed a fluorescent acetylcholine sensor (iAChSnFR) in the dorsal striatum of 11 mice and imaged it through an implanted lens with two-photon microscopy. Acetylcholine is a neurotransmitter with many jobs, from muscle control to attention and learning. In the striatum—a basal ganglia hub for action selection and habit formation—it comes mostly from a small population of cholinergic interneurons.

Three findings stand out:

Rewards produced dips. During initial learning, a delivered reward was followed by a brief drop in acetylcholine. Ordinary errors produced only smaller dips.
Unexpected non-reward after the reversal produced rises. When a mouse chose the formerly rewarded arm and got nothing, acetylcholine increased across most of the imaged tissue—about 85% of the small (~50 × 50 µm) patches the team analyzed.
The rise tracked what the mouse did next. Signals were larger on failed trials that preceded a switch to the other arm ("lose-shift") than on trials followed by a repeat. Across animals, larger responses correlated with a higher probability of lose-shift (r = 0.67, n = 11).

The responses were also spatially uneven. Patches showing increases sat more toward the medial striatum, with a trend toward anterior positions; patches showing decreases sat more laterally.

What the intervention showed. Correlation alone cannot show that acetylcholine mattered, so in a separate group of ten mice the researchers used a chemogenetic tool to suppress striatal cholinergic interneurons in half of them during reversal. Those mice made fewer lose-shift choices, performed worse in the first reversal session, and needed more trials to reach criterion (about 115 versus 63). Win-stay behavior and running speed were not significantly affected.

What remains interpretation. The inhibition was not time-locked to specific trials, so it shows these neurons are needed for timely adaptation in this task, not exactly when their signal acts. The authors themselves note they cannot yet separate a learning signal from an affective or motivational reaction to the missing reward. Their proposal—that spatially patchy acetylcholine might let new associations form in some striatal territory while sparing old ones elsewhere—is a hypothesis, not a result. And the whole study used a single reversal in a single task, with small groups.

What the study does give us is a striking contrast: the same outcome—no reward—was followed by opposite acetylcholine responses before and after a well-learned rule broke. The authors read this as a signal of violated expectation. Whatever its precise meaning, that distinction is the crux of the computational problem.

When Expectations Fail: The Computational View

Reinforcement learning formalizes expectation as a value estimate and surprise as a prediction error. A widely used version is the temporal-difference (TD) error from TD(0) learning, as presented in Sutton and Barto's textbook:

text
δt = r_t + γ · V(s{t+1}) − V(s_t)
δt — the temporal-difference error at time t
r_t — the immediate reward received after leaving state s_t
γ — the discount factor (0 ≤ γ ≤ 1), weighting future value
V(s_t) — the agent's estimated value of the current state
V(s{t+1}) — the estimated value of the next state

(Sutton and Barto write the reward as R_{t+1}; it is the same quantity under a different indexing convention.) This is one formulation among many. Q-learning, actor-critic methods, and policy-gradient methods use related but distinct error signals.

A worked example. Suppose an agent has learned that a choice leads to reward 90% of the time, so V(s_t) = 0.9. The trial then ends, so V(s_{t+1}) = 0. Let γ = 0.9.

If the reward arrives (r = 1): δ = 1 + 0.9 × 0 − 0.9 = +0.1
If the reward is missing (r = 0): δ = 0 + 0.9 × 0 − 0.9 = −0.9

With a learning rate of 0.1, the omission nudges the estimate from 0.9 to 0.81.

Now notice something uncomfortable. If the reward really does arrive 90% of the time, a −0.9 error appears on one trial in ten, and the errors average to zero. After a reversal, the identical −0.9 appears on the identical kind of trial. A single prediction error carries no label saying "bad luck" or "the rules changed." It drives a value update; it does not diagnose its own cause. Distinguishing noise from a genuine shift requires integrating evidence across trials and holding some model of how volatile the environment is likely to be. That is change detection, a problem related to—but separate from—prediction error.

This is where a careful reader should resist a tempting shortcut. Dopamine, not acetylcholine, is the neuromodulator classically linked to TD-style errors: Schultz, Dayan, and Montague (1997) reviewed evidence that the phasic firing of primate dopamine neurons resembles a reward prediction error. Striatal acetylcholine is not a copy of that signal; Duhne and colleagues (2024) found that, in rats, cholinergic pauses did not line up in time or location with dopamine prediction-error signals. Nothing in the mouse study shows acetylcholine computing the equation above. The equation is an educational bridge—a way of stating precisely what problem the mice were solving, not a description of how they solved it.

Exploration Versus Exploitation

Even an agent that suspects something has changed still has to decide what to do about it. Reinforcement learning frames this as a tradeoff:

Exploitation — keep taking the action that currently looks best.
Exploration — try alternatives that might reveal something better.

In a changing world, too little exploration anchors an agent to an outdated policy; too much throws away behavior that was still working.

The simplest strategy, epsilon-greedy selection, exploits most of the time and picks a random action with small probability ε. It is simple and gives every action a nonzero probability of being sampled, though no finite run guarantees that every action will be selected. But exploration alone does not make an agent adaptive; what it does with the information matters as much as gathering it.

The simulation below is the opening scenario in code: arm 0 pays out with probability 0.8 and arm 1 with 0.2, and the probabilities swap at step 500. Both agents use ε = 0.1 and differ only in how they average rewards. The first uses a sample average, so every reward in its history counts equally. The second uses a constant step size (α = 0.1), so recent rewards count more—a standard approach Sutton and Barto describe for tracking nonstationary problems.

python
"""Two-armed bandit whose reward probabilities reverse partway through.

Simulated agents only. Standard library, Python 3.8+.
Arm 0 pays 1 with p=0.8 and arm 1 with p=0.2, then they swap at step 500.
Both agents use epsilon-greedy selection (epsilon = 0.1). They differ in how
they update value estimates:
sample average : step size 1/n (every past reward weighted equally)
constant step size : alpha = 0.1 (recent rewards weighted more)
"""
import random

STEPS, SWITCH, RUNS, EPSILON = 1500, 500, 1000, 0.1
P_BEFORE, P_AFTER = [0.8, 0.2], [0.2, 0.8]

def run(rng, alpha=None):
q, n = [0.0, 0.0], [0, 0]
chose_best = []
for t in range(STEPS):
probs = P_BEFORE if t < SWITCH else P_AFTER
if rng.random() < EPSILON:
a = rng.randrange(2)
else:
a = 0 if q[0] > q[1] else 1 if q[1] > q[0] else rng.randrange(2)
r = 1.0 if rng.random() < probs[a] else 0.0
n[a] += 1
step = alpha if alpha is not None else 1 / n[a]
q[a] += step * (r - q[a]) # the bandit form of a prediction error update
chose_best.append(a == probs.index(max(probs)))
return chose_best

def summarize(label, alpha):
rng = random.Random(42) # fixed seed: identical numbers on every run
totals = [0] * STEPS
for _ in range(RUNS):
for t, ok in enumerate(run(rng, alpha)):
totals[t] += ok
rate = [x / RUNS for x in totals]
before = sum(rate[SWITCH - 100:SWITCH]) / 100
after = sum(rate[SWITCH:SWITCH + 200]) / 200
late = sum(rate[-100:]) / 100
first_crossing = next(t - SWITCH for t in range(SWITCH, STEPS) if rate[t] > 0.5)
print(f"{label:<20} best-arm rate | last 100 before switch: {before:.2f}"
f" | first 200 after: {after:.2f} | final 100: {late:.2f}"
f" | first step >50%: {first_crossing}")

summarize("sample average", None)
summarize("constant alpha=0.1", 0.1)

Simulated output (seed 42, 1,000 runs per agent). Each rate is the share of runs choosing the currently better arm, averaged over the stated window. "First step >50%" is the first post-switch step at which more than half of the runs chose the new best arm. It is an illustrative threshold crossing, not a per-agent recovery time or a measure of sustained adaptation:

Before the switch, the two agents perform almost identically by this metric. Afterward, the sample-average runs mostly stay on the old arm: only 5% of their choices in the first 200 post-switch steps go to the new best arm, and the majority does not cross over until step 519—even though every agent keeps exploring with the same probability throughout. Each agent's estimate for arm 0 is anchored by hundreds of earlier successes, so each new failure moves it only by 1/n. The constant-step runs cross the 50% line by step 21 and stay above it.

This is not a verdict that constant step sizes are better. They never fully converge, so in a truly stationary problem they keep reacting to noise. Neither agent detects the change; one simply forgets faster. More principled methods model the change explicitly—sliding-window and discounted UCB base their confidence estimates on discounted or recent observations, and Bayesian online change-point detection maintains a running probability over how long ago the last change occurred. Each embeds an assumption about how often the world shifts, and each pays for it when that assumption is wrong.

What Neuroscience Can—and Cannot—Teach AI Engineers

The legitimate connection between the mouse study and the bandit is a shared problem, not a shared mechanism. Both must answer the same questions. Was this failure noise or signal? How many failures justify reconsidering a learned strategy? When the strategy changes, should the old one be erased or kept in reserve?

The neuroscience offers useful prompts:

Context-dependent responses to failure. In the study, ordinary errors during learning and unexpected failures after a well-learned rule broke produced different acetylcholine signals. If the authors' reading is right, the response to a bad outcome depended on how established the expectation had been. A fixed ε has no such sensitivity; it explores at the same rate whether the agent is confident or confused.

Uncertainty as a control variable. Computational neuroscientists have long proposed that neuromodulators report different kinds of uncertainty. Yu and Dayan (2005) suggested that acetylcholine signals expected uncertainty and norepinephrine signals unexpected uncertainty, such as an unsignaled context switch. That theory concerned cortical circuits, and a reversal like the one in this study would arguably fall in its "unexpected" category—a reminder that mapping neuromodulators onto computational variables remains unsettled.

Adapting without erasing. The authors' hypothesis that patchy acetylcholine might protect old associations while new ones form echoes a problem continual-learning researchers know well: catastrophic forgetting. It is an intriguing parallel, but it is a hypothesis about mice, not evidence for any AI method.

The limits need stating just as plainly:

A single-reversal task in small groups of mice cannot establish how human decision-making works.
Artificial agents have no cholinergic interneurons; "neuromodulation" in a model is a design analogy.
Similar behavior does not imply similar internal processes.
The study proposes no AI algorithm.
Nothing in it shows that copying acetylcholine dynamics would improve agent performance.
Biology can inspire a computational hypothesis; only experiments on the computational system can validate it.

The Bigger Engineering Lesson

Return to the two-button agent. Before the switch, it looked excellent: 95% optimal choices. That number described its performance under the original reward structure, but it did not reveal how slowly the agent would adapt when that structure changed. Robustness to change shows up only when you test under change.

That is why optimizing against a fixed reward structure is not enough. Adaptive systems need a way to update expectations, an estimate of how much to trust those expectations, and a policy for responding when they fail. None of these settings is universally correct. Switching after every disappointment is just as maladaptive as never switching; the switching the mouse study examined followed failures that violated a well-established rule, not every missing reward.

The mice did not solve reinforcement learning. They did illustrate, with unusual precision, where the hard part lies.

Learning which decisions lead to reward is only half of intelligence. The other half is recognizing when yesterday's right decision has quietly become today's mistake.

References

Sarpong, G. A., Pass, R., Liyanagama, K., Chu, K.-Y., Kurima, K., Akamine, Y., Chouinard, J. A., Looger, L. L., & Wickens, J. R. (2025). Spatially heterogeneous acetylcholine dynamics in the striatum promote behavioral flexibility. Nature Communications, 16, 10877. https://doi.org/10.1038/s41467-025-66826-1

Sutton, R. S., & Barto, A. G. (2018). Reinforcement Learning: An Introduction (2nd ed.). MIT Press. http://incompleteideas.net/book/the-book-2nd.html

Schultz, W., Dayan, P., & Montague, P. R. (1997). A neural substrate of prediction and reward. Science, 275(5306), 1593–1599. https://doi.org/10.1126/science.275.5306.1593

Duhne, M., Mohebi, A., Kim, K., Pelattini, L., & Berke, J. D. (2024). A mismatch between striatal cholinergic pauses and dopaminergic reward prediction errors. Proceedings of the National Academy of Sciences, 121(41), e2410828121. https://doi.org/10.1073/pnas.2410828121

Yu, A. J., & Dayan, P. (2005). Uncertainty, neuromodulation, and attention. Neuron, 46(4), 681–692. https://doi.org/10.1016/j.neuron.2005.04.026

Garivier, A., & Moulines, E. (2011). On upper-confidence bound policies for switching bandit problems. In Algorithmic Learning Theory (ALT 2011). Springer. https://doi.org/10.1007/978-3-642-24412-4_16

Adams, R. P., & MacKay, D. J. C. (2007). Bayesian online changepoint detection. arXiv:0710.3742. https://arxiv.org/abs/0710.3742

Okinawa Institute of Science and Technology Graduate University. (2026, June 8). Scientists discover the brain chemical that helps you break bad habits. ScienceDaily. https://www.sciencedaily.com/releases/2026/06/260606075901.htm

Your Wi-Fi Has No Walls: What Aluminum Foil Reveals About Wireless Security

Khali Sollis — Tue, 13 Oct 2026 18:00:00 +0000

How a 2017 experiment challenges the way we think about network boundaries, signal propagation, and defense in depth.

The Invisible Network Beyond Your Walls

A home has walls, doors, and locks. A wireless network does not necessarily stop at any of them. Its edge is set by physics rather than property lines: a 2.4 GHz Wi-Fi signal has a wavelength of roughly 12.5 cm, and how much of it survives a wall depends on the frequency, the material, its thickness, the angle of incidence, and everything else in the room. Drywall and wood let a good deal through. Concrete, brick, and metal take more. The result is a coverage footprint shaped like no floor plan.

That spillover is a fact about propagation, not automatically a vulnerability. A beacon that is detectable on the sidewalk only matters if someone on the sidewalk can do something useful with it. Still, the mismatch between the physical boundary of a home and the radio boundary of its network raises a fair engineering question: if we could shape where the signal goes, would that add a meaningful layer of protection?

The 2017 Experiment Behind the 2026 Headline

A news story circulating in 2026 revived this question, but the research behind it is nine years old. In November 2017, Xi Xiong, Justin Chan, Ethan Yu, Nisha Kumari, Ardalan Amiri Sani, Changxi Zheng, and Xia Zhou presented "Customizing Indoor Wireless Coverage via 3D-Fabricated Reflectors" at ACM BuildSys in Delft. It extended a 2015 workshop paper by Chan, Zheng, and Zhou that introduced the system's name, WiPrint.

The approach is computational, not improvised. The user supplies a coarse 3D model of the space (walls and doors, no furniture), the access point's location and antenna positions, and a set of target areas where signal should be stronger or weaker. A 3D ray tracer, accounting for reflection, transmission, and diffraction, predicts signal strength in each 1 m × 1 m cell. The reflector surface is represented as a smooth parametric NURBS surface, and a simulated-annealing search guided by gradient descent reshapes it until predicted coverage best matches the target. On a 2017 laptop this took about 23 minutes. The shape is then 3D-printed in plastic, roughly 20 × 20 cm for under $35, coated in 0.024 mm household aluminum foil, and mounted around the router's antennas.

The team tested this in two furnished indoor scenes, a 19 × 13 m workspace and a 16 × 12 m home-like layout, with people moving around. The main access point was a Netgear R7000 (802.11ac, three antennas) set to 10 mW and measured on the 2.4 GHz band; two older 2.4 GHz routers were also tested. A programmed drone hovered at each location to log signal strength, and laptops running iperf measured throughput.

The widely repeated numbers check out, with a qualifier: the paper reports that optimized reflectors weakened signals in target areas by up to 10 dB and strengthened them by up to 6 dB. These are best-case figures in specific target regions, not averages across a home. In the single-router test, 83.1% of cells moved in the intended direction; in the two-router test, 91%. Throughput changes ranged from −63.3% to +55.1% across seven of nine sampled locations in the two-router scene. A naive concave reflector, the soda-can approach, simply weakened everything behind it.

Why Aluminum Reflects Wi-Fi Signals

Wi-Fi is electromagnetic radiation. When it meets a good conductor, free electrons in the metal respond to the oscillating field and re-radiate, so most of the incident energy is reflected rather than passed through. At 2.4 GHz, current in aluminum is confined to a skin depth of under two micrometers, which is why even kitchen foil is electrically thick enough to act as a reflector. The researchers found copper and aluminum performed similarly, silver slightly better, and thicker sheets improved blocking only moderately.

Energy striking the reflector is split between reflection, absorption, and some transmission. The authors observed that their reflectors were better at weakening than strengthening, and the reason is instructive: only reflected energy can boost the signal in front of the reflector, while both reflected and absorbed energy reduce it behind. A passive reflector adds no power; it only redirects what the antennas already radiate. Sending more energy toward one area generally leaves less for others, and because the redirected waves also interfere with the room's existing multipath, the reflector can shift signal strength in places it was never aimed at, upward in some spots and downward in others. The paper documents one such side effect: weakening the hallway also slightly weakened part of the lab.

The flashlight-mirror analogy helps up to a point and then fails. A 20 cm reflector is less than two wavelengths across at 2.4 GHz, so waves diffract around its edges instead of being cleanly redirected. Indoors, signals also arrive by many paths at once, bouncing off walls, floors, and furniture. This multipath propagation means the copies combine constructively in some spots and destructively in others, so readings can change by several decibels over a few centimeters. That is why a shape optimized for one floor plan is the wrong shape for another.

What the Experiment Demonstrated—and What It Didn't

What the study supports is specific: under its conditions, an optimized, carefully placed reflector redistributed received signal strength and throughput by moderate amounts in the intended direction across most cells. The effect tolerated about 10° of orientation error and 10 cm of placement error.

What it did not establish matters as much. All signal maps were collected at 2.4 GHz, so nothing in the paper speaks directly to 5 GHz, 6 GHz, Wi-Fi 6, 6E, or 7, wide channels, or modern routers with internal antenna arrays and beamforming. The authors themselves describe the RSS and throughput changes as moderate and attribute the limit to Wi-Fi's wavelength. Nor does the paper show that a hand-shaped sheet of foil behind a router reproduces the result; its own naive reflector did not.

Outcomes will also vary with antenna configuration, room layout, building materials, client placement, moving people, and neighboring networks. And received signal strength is not throughput, reliability, or security. Even the model's predictions carried a mean error of about 3 dBm.

Signal Containment Is Not Encryption

Coverage control answers where a signal can be received. Authentication answers who may join the network. Encryption answers what someone who receives the signal can learn. A reflector addresses only the first.

The paper's security test measured packet loss at locations outside each router's intended region. At some locations that could previously reach the network, packet loss rose to 50–60%, enough that the authors treat the network as inaccessible there. But at several others, including one close to the router, packet loss barely changed, and the authors proposed adding transmit-power control to address it. That is evidence of reduced usability at specific points for ordinary laptops. It is not evidence about an attacker.

A lower received signal level, often reported through RSSI (received signal strength indicator) or a device-specific signal measurement, may make reception more difficult without making it impossible. An adversary can move closer, wait outside a different wall, or use a directional antenna with substantially more gain than a laptop's. The researchers themselves frame physical confinement as complementary to encryption, and that is the defensible framing.

The protections that actually govern access and confidentiality live higher up, and each covers a different stretch of the path:

WPA2/WPA3 authenticate devices joining the network and encrypt traffic over the air between each device and the access point. Their protection ends at the router. WPA3-Personal replaces WPA2's pre-shared key handshake with SAE, which the Wi-Fi Alliance says gives users stronger protection against password-guessing attempts; WPA3 networks also require Protected Management Frames and disallow outdated protocols.
HTTPS/TLS encrypts and authenticates the connection between an application and the specific server it talks to, across every network in between. It protects content whether or not the Wi-Fi is trustworthy, though observers can still see which hosts you contact.
VPNs wrap traffic in an encrypted tunnel from your device to a VPN server, hiding traffic routed through the tunnel from the local network and your ISP. Beyond that server, traffic is only as protected as its own encryption, and you are trusting the VPN operator. A VPN can be useful for accessing corporate networks or adding privacy against local network observers, but it does not replace HTTPS and is not automatically necessary on public or home Wi-Fi.

The NSA's home network guidance translates this into practical steps:

Use WPA3, or WPA2/WPA3 mode only while older devices still need it, and enable Protected Management Frames where available.
Choose a passphrase of at least twenty characters.
Keep router firmware updated, and replace routers that no longer receive support.
Protect the router's admin login with strong authentication, and disable remote administration and UPnP.
Segment primary, guest, and IoT devices onto separate networks.

This is defense in depth: independent layers, each with a clearly stated job, so that one failure does not become total failure. For organizations, NIST SP 800-153 applies the same thinking to WLAN configuration and monitoring.

How Developers Can Investigate Their Own Wireless Coverage

A controlled comparison at home teaches a lot about propagation. It will not validate foil as a security control.

Pick five to ten fixed measurement spots, including a few near exterior walls, and mark them.
Record a baseline: at each spot, take at least ten readings with the same device, held the same way.
Place a reflector behind the router without covering vents or touching power connections.
Repeat the same readings at the same spots.
Compare medians and spread, not single readings.

Keep the router, client device, band, channel, and orientation constant. On Linux, iw dev link reports signal in dBm; Windows' netsh wlan show interfaces reports a percentage, which is not directly comparable. RSSI scales are vendor-defined and uncalibrated, so trust differences on one device more than absolute numbers across devices.

A decibel describes a ratio: ΔP(dB) = 10 · log₁₀(P₂ / P₁). dBm is an absolute level referenced to one milliwatt, so −60 dBm is one nanowatt. Subtracting two dBm readings gives a difference in dB. A 10 dB drop means one-tenth the received power, not one-tenth the range: in free space, where power falls with the square of distance, it corresponds to roughly a 3.2× change in distance, and indoors the relationship depends on the environment. It says nothing direct about throughput or attack probability.

This script compares the median reading at each spot before and after, and reports the spread of each set. The sample data is hypothetical, ten readings per spot to match the procedure above, and two spots stand in for your full list.

python
import statistics as st
import matplotlib.pyplot as plt

HYPOTHETICAL readings in dBm — replace with your own measurements

baseline = {
"desk": [-48, -50, -49, -47, -51, -49, -48, -50, -52, -49],
"porch": [-71, -74, -72, -73, -70, -72, -75, -71, -73, -72],
}
modified = {
"desk": [-45, -46, -44, -47, -45, -46, -44, -45, -47, -46],
"porch": [-78, -80, -77, -79, -81, -78, -80, -79, -77, -82],
}

locations = list(baseline)
deltas = [st.median(modified[k]) - st.median(baseline[k]) for k in locations]

for k, d in zip(locations, deltas):
before = max(baseline[k]) - min(baseline[k])
after = max(modified[k]) - min(modified[k])
print(f"{k:>6}: median change {d:+.1f} dB (power ×{10 ** (d / 10):.2f}), "
f"spread {before} dB before, {after} dB after")

plt.bar(locations, deltas, color=["tab:green" if d > 0 else "tab:red" for d in deltas])
plt.axhline(0, color="black", linewidth=0.8)
plt.ylabel("Median change (dB)")
plt.title("Change in median RSSI per spot (hypothetical data)")
plt.show()

The bar chart shows the change in median RSSI at each named spot. It is not a coverage heatmap; for that, you would plot the same deltas at each spot's position on a floor plan.

Test only networks you own or are authorized to measure, never block router ventilation, and keep foil away from power supplies and outlets.

The Bigger Engineering Lesson

Software engineers work mostly above the physical layer, for good reason. But coverage is a design parameter, shaped by geometry and materials as much as by configuration files, and WiPrint makes that unusually visible.

The lesson is not that foil secures Wi-Fi. It is that every security control should come with an honest statement of what it protects against and what evidence supports it. Signal shaping can modestly reduce where a network is usable, under measured conditions, at 2.4 GHz. Encryption, authentication, and secure configuration protect what the network carries and who may join. The two belong in the same threat model, at different layers, and neither substitutes for the other.

We put enormous effort into protecting the information that travels through our networks. Knowing where the network itself physically travels belongs in that same engineering conversation.

References and Further Reading

Xiong, X., Chan, J., Yu, E., Kumari, N., Amiri Sani, A., Zheng, C., & Zhou, X. (2017). Customizing Indoor Wireless Coverage via 3D-Fabricated Reflectors. Proceedings of the 4th ACM International Conference on Systems for Energy-Efficient Built Environments (BuildSys '17). https://doi.org/10.1145/3137133.3137148 · PDF

Chan, J., Zheng, C., & Zhou, X. (2015). 3D Printing Your Wireless Coverage. ACM Workshop on Hot Topics in Wireless (HotWireless). https://cs.dartmouth.edu/~xia/papers/hotw15-wiprint.pdf

DartNets Lab, Dartmouth College. WiPrint project page. https://dartnets.cs.dartmouth.edu/wiprint

Sheth, A., Seshan, S., & Wetherall, D. (2009). Geo-fencing: Confining Wi-Fi Coverage to Physical Boundaries. Pervasive Computing (Pervasive 2009), 274–290. Springer. https://doi.org/10.1007/978-3-642-01516-8_19

Souppaya, M., & Scarfone, K. (2012). Guidelines for Securing Wireless Local Area Networks (WLANs), NIST SP 800-153. National Institute of Standards and Technology. https://doi.org/10.6028/NIST.SP.800-153

Wi-Fi Alliance. Security: WPA3, Wi-Fi Enhanced Open, and Protected Management Frames. https://www.wi-fi.org/security

National Security Agency (2023). Best Practices for Securing Your Home Network, Cybersecurity Information Sheet. https://media.defense.gov/2023/Feb/22/2003165170/-1/-1/0/CSI_BEST_PRACTICS_FOR_SECURING_YOUR_HOME_NETWORK.PDF

Dependency Injection: How Some People Quietly Program Themselves Into Your Identity

Khali Sollis — Fri, 25 Sep 2026 12:00:00 +0000

How emotional dependency becomes embedded inside self-concept, why certain relationships begin influencing your decisions from the inside, and what happens when another person slowly becomes part of your psychological operating system

Some relationships affect you externally.

Others go much deeper.

They quietly integrate themselves into your internal architecture.

At first, it feels harmless:

considering their opinion
adapting to their preferences
emotionally consulting them internally
imagining their reaction before making decisions

Over time, something more subtle happens.

You stop simply loving the person.

You start organizing yourself around them.

That was the realization that changed everything for me:

Certain people do not just enter your life.

They become embedded inside your identity infrastructure.

The Bug: Dependency Injection

In software architecture, dependency injection occurs when external components become integrated into a system’s internal functioning.

The system begins relying on those components to operate normally.

Emotionally, the same thing happens when another person becomes psychologically embedded into:

decision-making
self-worth
emotional regulation
identity stability
future orientation
internal validation systems

The self slowly stops operating independently.

The Integration Happens Gradually

This process rarely feels dangerous at first.

In fact, it often feels like intimacy.

You begin:

thinking through their perspective automatically
adjusting behavior around their preferences
filtering choices through their approval
imagining life outcomes based on their presence

Eventually, their psychological presence remains active even when they are physically absent.

That is where dependency deepens.

The Person Becomes an Internal Reference Point

One uncomfortable truth:

At some point, I stopped making certain decisions based on myself.

Instead, my mind automatically evaluated:

How would they react?
Would they approve?
Would this threaten the connection?
Would they still choose me?

Without realizing it, another person had become part of my internal operating logic.

Dependency Often Masquerades as Love

This distinction became critical.

Love:

allows connection while preserving selfhood

Dependency:

merges emotional stability with another person’s presence

One expands life.

The other quietly restructures identity around emotional access.

Why Dependency Injection Feels So Powerful

Because the nervous system associates attachment with safety.

Especially when someone becomes tied to:

emotional relief
belonging
identity reinforcement
validation
nervous system regulation

The brain begins integrating the person into survival processing.

Their presence starts feeling psychologically necessary.

The Internalization Process

The progression often looks like this:

connection
→ emotional reliance
→ behavioral adaptation
→ identity integration
→ dependency stabilization

The dangerous part is that the process feels emotionally meaningful while it is happening.

Sometimes even romantic.

The Self Slowly Becomes Negotiated

Dependency changes internal behavior quietly.

You start:

abandoning preferences
minimizing needs
reshaping routines
adjusting goals
suppressing authenticity

not because someone explicitly demanded it…

but because preserving the attachment becomes psychologically prioritized.

Why Separation Feels So Disorienting

Because dependency removal feels like identity disruption.

When someone deeply embeds into your internal systems, their absence affects:

emotional regulation
daily habits
future imagination
self-perception
nervous system stability

People often think they are grieving only the relationship.

Sometimes they are grieving the collapse of the psychological structure built around it.

The Difference Between Bonding and Psychological Merging

Healthy bonding:

creates closeness while maintaining autonomy

Dependency injection:

blurs the boundary between connection and identity

One preserves individuality.

The other slowly absorbs it.

The Most Dangerous Part

The most dangerous part is not emotional pain.

It is self-disconnection.

Eventually you stop asking:

What do I want?

And start asking:

What preserves the emotional system built around them?

That shift changes your entire relationship with yourself.

The Realization

Eventually, I stopped asking:

Why does losing this connection feel like losing myself?

And started asking:

How much of my identity became structurally dependent on this person’s presence?

That question exposed the architecture underneath the attachment completely.

The Fix: Dependency Extraction

Healing required more than emotional distance.

It required rebuilding independent internal systems.

Not coldness.

Autonomy.

Rebuild Internal Decision-Making

I began practicing making choices without emotionally consulting:

their approval
their reaction
their preferences
their imagined perception

At first, it felt unnatural.

That discomfort revealed how deep the dependency had become.

Separate Love From Psychological Reliance

Old model:

If someone matters deeply to me, my identity should naturally revolve around them.

Updated model:

Healthy connection supports identity. It should not replace it.

That distinction restored clarity.

Reclaim Internal Territory

This part was critical.

I started reclaiming:

routines
goals
emotional regulation
creative focus
future planning

without organizing them around another person’s presence.

The nervous system slowly relearned self-sufficiency.

Allow Relationships to Be Additions, Not Operating Systems

This became the new framework:

People can matter deeply without becoming:

your emotional infrastructure
your identity anchor
your source of self-worth
your primary nervous system regulator

Connection is healthiest when it enhances identity instead of consuming it.

What Changed

After identifying dependency injection patterns:

emotional autonomy increased
self-trust strengthened
relationships became clearer
identity stability improved
fear of abandonment decreased significantly

And most importantly:

I stopped confusing emotional merging with emotional depth.

Reframing Attachment

Old model:

The deeper the attachment, the more central they should become to my identity.

Updated model:

Healthy attachment preserves individuality while allowing connection.

That distinction restored balance.

Closing the Series

This entire series began with one realization:

Many emotional struggles are not random.

They are systems.

Patterns.
Conditioning.
Behavioral loops.
Identity distortions.
Nervous system adaptations.

For years, I believed emotional exhaustion was simply part of being “good,” “loving,” “understanding,” or “available.”

It wasn’t.

Many of those behaviors were unexamined survival strategies running automatically underneath my personality.

This series was never about becoming cold.

It was about identifying the hidden architectures that quietly teach people to:

abandon themselves
over-function emotionally
confuse instability with love
mistake exhaustion for devotion
tolerate dynamics that slowly erase them

Awareness changes the system.

Not instantly.

But permanently.

Final Takeaway

Some people do not merely influence your emotions.

They quietly integrate themselves into your identity until your internal stability becomes partially dependent on their existence.

Healing begins when you reclaim authorship over your own psychological architecture.

Because relationships should complement your identity.

Not become the operating system running it.

Final Status
Behavioral anti-pattern audit: completed
Dependency injection vulnerabilities: identified
Self-authorship systems: restoring
Internal architecture ownership: returned
Series Complete: Behavioral Anti-Patterns
The Audit of the “Nice” Trap: A Documentation of Self-Erasure
Missing Boundary Checks: Why “Nice” Code Always Gets Exploited
Unbounded Processes: The Hidden Cost of Always Saying Yes
Deferred Execution: Why Avoiding “No” Creates Worse Outcomes
Infinite Approval Loops: Breaking the Need for External Validation
Caching Bad Data: Why We Keep Trusting Broken Patterns
Handling Exceptions That Aren’t Yours: The Trap of Emotional Over-Responsibility
Refactoring Identity: When “Being Nice” Becomes Technical Debt
No Rate Limiting: How Instant Availability Destroys Focus
Broken Access Control: Who Actually Deserves Your Time?
From Approval-Seeking to System Integrity: Shipping a New Version of Yourself
Recursive Loneliness: Why Some Attachments Feel Impossible to Exit
Emotional Cache Invalidation: Accepting That Someone Has Changed
Ghost Dependencies: The Relationships That Still Control You After They End
Phantom Notifications: Why Your Nervous System Still Waits for People Who Left
Emotional Memory Leaks: Why Old Relationships Keep Reappearing During Vulnerable Moments
Emotional Rollbacks: Why We Revert to Old Versions of Ourselves Around Certain People
Identity Drift: How We Slowly Abandon Ourselves to Keep Connections Alive
Approval Latency: Why Waiting for Validation Delays Your Entire Life
Emotional Deadlocks: When Two People Wait for Each Other to Change First
Dependency Injection: How Some People Quietly Program Themselves Into Your Identity

Emotional Deadlocks: When Two People Wait for Each Other to Change First

Khali Sollis — Fri, 18 Sep 2026 13:00:00 +0000

How unresolved relational dynamics create emotional standstills, why mutual avoidance freezes connection, and what happens when two people become trapped in passive psychological negotiations

Some relationships do not end dramatically.

They stall.

Communication weakens.
Resentment accumulates.
Distance increases.
But neither person fully leaves.

Instead, both people enter a silent negotiation:

I will change once they change first.

And the relationship slowly freezes there.

No repair.
No resolution.
No honest rupture.

Just emotional deadlock.

The Bug: Emotional Deadlocks

In computing systems, a deadlock occurs when two processes become permanently blocked because each is waiting for the other to act first.

Neither process progresses.

Human relationships can operate the same way.

Examples:

both people wait for the other to apologize
both wait for vulnerability first
both wait for emotional maturity first
both wait for effort first
both wait for accountability first

The connection becomes trapped inside mutual hesitation.

Why Deadlocks Feel So Draining

Because emotional energy remains allocated without movement.

The system never fully:

repairs
resolves
reconnects
separates

Which creates continuous psychological occupation.

Part of the mind stays stuck monitoring:

possibility
disappointment
unresolved tension
imagined future resolution

The relationship remains emotionally open without actually functioning.

Passive Waiting Masquerades as Hope

This realization was uncomfortable.

Sometimes what people call:

hope

is actually:

prolonged emotional avoidance

Both people sense the dysfunction.

But instead of confronting reality directly, they remain psychologically suspended waiting for the other person to become different first.

Mutual Fear Creates Stagnation

Deadlocks are usually powered by fear.

Fear of:

rejection
vulnerability
accountability
abandonment
emotional exposure
losing control of the dynamic

So both people unconsciously choose emotional paralysis over uncomfortable clarity.

The Emotional Negotiation Loop

The pattern often looks like this:

Person A withdraws
→ Person B becomes defensive
→ Person A waits for accountability
→ Person B waits for warmth
→ neither moves
→ resentment compounds

Over time, inactivity itself becomes the system architecture.

Why Deadlocks Often Last for Years

Because ambiguity allows emotional fantasy to survive.

As long as the relationship remains unresolved:

hope remains technically possible
emotional attachment stays partially active
accountability stays avoidable

Deadlocks preserve potential while postponing reality.

The Difference Between Patience and Stagnation

This distinction changed everything.

Patience:

supports healthy progress

Deadlock:

suspends necessary movement indefinitely

One creates space for growth.

The other delays truth.

Emotional Pride Intensifies the Freeze

Sometimes neither person wants to “lose” emotionally.

Which means:

nobody apologizes first
nobody communicates honestly first
nobody softens first
nobody leaves first

The relationship becomes a psychological standoff disguised as emotional complexity.

Why Some People Prefer Deadlocks Over Endings

Because endings create certainty.

Deadlocks preserve illusion.

An unresolved relationship still allows fantasies like:

“Maybe later.”
“Maybe they’ll realize.”
“Maybe timing is the issue.”
“Maybe things will shift naturally.”

The nervous system often prefers suspended hope over definitive grief.

Even when the suspension itself becomes painful.

The Hidden Cost of Emotional Stalemates

Deadlocks quietly consume:

attention
emotional energy
identity stability
future openness
nervous system capacity

Part of the self remains permanently allocated toward unresolved emotional processing.

That fragmentation affects everything else.

The Realization

Eventually, I stopped asking:

Why does this relationship feel impossible to resolve?

And started asking:

What truth are both people avoiding by waiting for the other to change first?

That question exposed the actual architecture underneath the stagnation.

The Fix: Breaking Emotional Deadlocks

Resolution required movement.

Not perfect mutual readiness.

Movement.

Because deadlocks survive through indefinite hesitation.

Accept That Waiting Is Also a Decision

This part was critical.

Inaction is not neutral.

Every day spent inside unresolved dynamics reinforces the structure itself.

Deadlocks continue because both people continuously participate in maintaining them.

Stop Interpreting Silence as Emotional Depth

Old model:

This unresolved tension must mean the connection is profound.

Updated model:

Unresolved tension can also indicate avoidance, fear, and emotional immobility.

That distinction restored clarity.

Move Independently of Their Timeline

One powerful realization:

Your healing cannot depend on mutual synchronization.

If:

communication is absent
accountability is absent
effort is absent
growth is absent

you still retain responsibility for your own movement.

Waiting indefinitely for another person’s transformation delays your own life.

Prioritize Clarity Over Emotional Suspension

This became the new standard.

Not:

Does this connection still feel emotionally charged?

But:

Is this dynamic actually moving toward health, honesty, and reciprocity?

That question changes everything.

What Changed

After identifying emotional deadlock patterns:

unresolved dynamics became easier to recognize
emotional stagnation lost some of its romantic appeal
decision-making improved
self-respect strengthened
emotional energy returned

And most importantly:

I stopped confusing suspended movement with meaningful connection.

Reframing Waiting

Old model:

If I wait long enough, eventually they will become who I need them to be.

Updated model:

Mutual growth requires active participation, not indefinite emotional waiting.

That distinction restored momentum.

Takeaway

Emotional deadlocks occur when two people remain psychologically connected while simultaneously waiting for the other person to change first.

The result is emotional suspension:

unresolved tension
prolonged ambiguity
stalled growth
fragmented attention

Healing begins when clarity becomes more important than preserving possibility.

Because some relationships do not fail from lack of feeling.

They fail from lack of movement.

Status
Emotional deadlock detection: active
Passive waiting behaviors: identified
Forward movement protocols: restoring
Series: Behavioral Anti-Patterns

Previous: Approval Latency: Why Waiting for Validation Delays Your Entire Life
Next: Dependency Injection: How Some People Quietly Program Themselves Into Your Identity

Approval Latency: Why Waiting for Validation Delays Your Entire Life

Khali Sollis — Fri, 11 Sep 2026 13:00:00 +0000

How external validation creates emotional bottlenecks, why approval-seeking slows personal evolution, and what happens when your decisions remain permanently queued behind other people’s opinions

One of the most expensive realizations of my life was understanding how much time I lost waiting.
Waiting for:

permission

reassurance

approval

agreement

recognition

emotional confirmation

Not always consciously.
Sometimes the waiting was subtle.
I would hesitate before decisions.
Overanalyze before speaking.
Delay action until someone else validated the direction first.
At the time, I called it:
being thoughtful
Looking back, it was something else entirely.
My life was running on approval latency.

The Bug: Approval Latency
In technology systems, latency refers to delays between request and response.
High latency slows the entire system.
Emotionally, approval latency happens when:
internal decisions
cannot fully execute until:
external validation
is received.
The result:

delayed action

delayed boundaries

delayed self-trust

delayed identity formation

delayed life progression

The system becomes externally bottlenecked.

The Hidden Dependency
This realization was uncomfortable:
Many of my decisions were not actually autonomous.
They were socially processed.
Internally, the sequence looked like this:
desire→ self-doubt→ external consultation→ emotional waiting→ permission received→ action finally begins
The problem was not reflection.
The problem was dependency.

Why Approval Feels Necessary
Because validation temporarily reduces uncertainty.
Especially for people conditioned to associate approval with:

safety

belonging

worthiness

emotional stability

acceptance

The nervous system learns:
approval=reduced emotional risk
So independent action begins feeling psychologically dangerous.

The Cost of Chronic Permission-Seeking
Approval latency quietly delays entire lives.
Not dramatically.
Incrementally.
You postpone:

projects

conversations

boundaries

decisions

reinvention

visibility

waiting for:

consensus

reassurance

emotional certainty

guaranteed acceptance

Meanwhile, life continues moving.

The Validation Queue Never Ends
One dangerous truth:
External approval systems are infinite.
No matter what you do:

someone will disagree

someone will misunderstand

someone will disapprove

someone will withdraw support

If your nervous system requires universal validation before action, execution becomes impossible.

Approval-Seeking Creates Identity Fragmentation
This part changed everything.
The more I optimized for approval, the less clear my actual preferences became.
Because constant external calibration weakens internal reference systems.
You stop asking:
What do I think?
And start asking:
What response minimizes rejection?
That shift slowly disconnects people from themselves.

Why Highly Sensitive People Often Struggle Here
Because emotionally perceptive people detect reactions quickly.
Micro-signals:

tone changes

hesitation

facial expressions

silence

disapproval cues

become emotionally significant.
The nervous system starts adapting behavior preemptively to avoid friction.
Approval latency increases.

The Emotional Economics of Delayed Action
Waiting for validation creates compounding costs.
Every delayed decision drains:

momentum

confidence

clarity

energy

self-trust

The longer action remains externally dependent, the weaker internal authority becomes.

The Difference Between Feedback and Permission
This distinction became critical.
Healthy feedback:
informs decisions
Approval dependency:
controls decisions
One expands intelligence.
The other weakens autonomy.

The Realization
Eventually, I stopped asking:
Why do I overthink everything before acting?
And started asking:
Why does my nervous system treat disapproval like danger?
That question exposed the actual architecture underneath the hesitation.

The Fix: Reducing Approval Dependency
Healing required strengthening internal execution systems.
Not becoming reckless.
Becoming self-authorized.

Shorten the Delay Between Knowing and Acting
I began noticing how often I already knew the answer internally before seeking external confirmation.
The issue was not lack of clarity.
It was lack of self-trust.
Reducing approval latency meant acting closer to initial internal knowing.
Normalize Discomfort During Independent Decisions
Old model:
If I feel anxious making this decision alone, it must be wrong.
Updated model:
Independent action often feels uncomfortable while self-trust is rebuilding.
That distinction changed execution completely.
Stop Outsourcing Identity
Some questions cannot be solved through consensus.
Especially:

values

boundaries

life direction

self-respect decisions

emotional standards

Too much external input creates internal noise.

Allow Disapproval Without System Failure This part was transformative. I started practicing:

saying things people may not like

making decisions without unanimous support

disappointing others without self-collapse

remaining stable during disagreement

The nervous system slowly learned:
disapproval≠emotional catastrophe
That lesson restored enormous freedom.

What Changed
After reducing approval latency:

decisions accelerated

overthinking decreased

self-trust strengthened

emotional dependence weakened

identity clarity improved

And most importantly:
I stopped placing my life on hold waiting for emotional authorization from other people.

Reframing Validation
Old model:
Once people approve, I can move forward confidently.
Updated model:
Confidence is built through aligned action, not universal approval.
That distinction restored momentum.

Takeaway
Approval latency delays lives quietly.
Not because people lack intelligence or capability, but because the nervous system becomes conditioned to treat external validation as a prerequisite for movement.
The result is chronic hesitation, fragmented identity, and delayed execution.
Healing begins when internal alignment becomes more important than external permission.
Because waiting for universal approval is one of the slowest ways to experience your own life.

Status
Approval dependency detection: activeExternal validation bottlenecks: reducingInternal execution systems: stabilizing

Series: Behavioral Anti-Patterns
Previous: Identity Drift: How We Slowly Abandon Ourselves to Keep Connections Alive
Next: Emotional Deadlocks: When Two People Wait for Each Other to Change First

Identity Drift: How We Slowly Abandon Ourselves to Keep Connections Alive

Khali Sollis — Fri, 04 Sep 2026 12:00:00 +0000

How self-erasure happens gradually inside relationships, why emotional adaptation can become identity loss, and what happens when maintaining connection becomes more important than maintaining yourself

Identity drift rarely happens dramatically.

Most people do not wake up one morning and consciously decide:

I am going to abandon myself for this relationship.

It happens slowly.

Quietly.

Incrementally.

A small compromise here.
A swallowed boundary there.
A personality adjustment that initially feels harmless.

Until one day you realize:

your behavior changed
your standards changed
your emotional tolerance changed
your priorities changed
your sense of self became unrecognizable

Not because you consciously chose self-erasure.

Because the connection slowly trained you to disconnect from yourself in order to preserve it.

The Bug: Identity Drift

In technology systems, identity drift occurs when a system gradually deviates from its original configuration over time.

Small unauthorized changes accumulate until the system no longer resembles its intended architecture.

Human relationships can create the same phenomenon.

Especially when emotional survival becomes tied to:

approval
acceptance
validation
emotional access
fear of abandonment

The self slowly adapts to maintain relational stability.

Even at its own expense.

The Drift Is Usually Rewarded at First

This was difficult to admit.

Many forms of self-abandonment initially receive positive reinforcement.

You become:

easier to deal with
more accommodating
more emotionally available
more forgiving
less confrontational

People often reward the version of you that requires the least resistance from them.

Which means identity drift can feel temporarily successful.

Small Self-Abandonments Compound

Identity erosion rarely begins with major sacrifices.

It starts microscopically.

Examples:

laughing at things that hurt you
tolerating disrespect to avoid conflict
pretending not to care
saying yes when you mean no
reducing your needs to remain “easy”
overexplaining your boundaries until they disappear

Each individual compromise seems manageable.

But accumulated compromises restructure identity over time.

The Nervous System Prioritizes Attachment

One uncomfortable truth:

Humans are biologically wired to preserve connection.

Especially if abandonment, rejection, or instability became emotionally significant earlier in life.

Which means the nervous system may unconsciously conclude:

maintaining attachment

maintaining authenticity

Once that equation activates, identity drift accelerates.

Adaptation Becomes Performance

At some point, I realized I was no longer relating naturally.

I was managing perception.

Monitoring:

tone
reactions
timing
emotional responses
approval signals

Parts of my personality became optimized for relational stability instead of internal truth.

That is not connection.

That is behavioral performance.

Why Identity Drift Feels Invisible

Because the changes happen gradually.

The system normalizes them incrementally.

You tell yourself:

“It’s not a big deal.”
“I’m just being understanding.”
“I’m trying to make things work.”
“Relationships require compromise.”

And they do.

But compromise becomes self-erasure when:

the relationship survives
while your authenticity slowly disappears
The Loss of Internal Reference

The most dangerous part of identity drift is not external.

It is internal disorientation.

Eventually you stop asking:

What do I actually want?

And start asking:

What response keeps this connection stable?

That shift changes everything.

Because your identity slowly becomes externally calibrated.

Emotional Shape-Shifting

Some relationships unconsciously reward emotional shape-shifting.

You become:

softer to avoid conflict
quieter to avoid rejection
more agreeable to maintain closeness
more available to avoid abandonment

Over time, adaptability becomes identity fragmentation.

You stop experiencing yourself consistently.

Why Exhaustion Appears

Maintaining distorted versions of yourself consumes enormous energy.

Because self-suppression requires continuous monitoring.

The nervous system remains hyper-alert:

adjusting
filtering
editing
minimizing
predicting reactions

The body eventually experiences this as emotional exhaustion.

Not because connection is hard.

Because performance is.

The Difference Between Growth and Self-Abandonment

This distinction became critical.

Healthy growth expands authenticity.

Identity drift suppresses it.

Growth says:

I am evolving into a more aligned version of myself.

Identity drift says:

I am becoming easier to tolerate so the connection survives.

Those are entirely different processes.

The Realization

Eventually, I stopped asking:

Why do I feel disconnected from myself lately?

And started asking:

What parts of myself have I been slowly negotiating away to maintain this connection?

That question exposed the architecture underneath the exhaustion.

The Fix: Identity Realignment

Recovery required rebuilding internal reference points.

Not through isolation.

Through honesty.

Notice Where You Self-Edit Automatically

The first signs were subtle:

hiding emotions
minimizing needs
avoiding honesty
suppressing discomfort
performing emotional neutrality

Those behaviors became indicators of drift.

Not maturity.

Stop Treating Self-Abandonment as Kindness

Old model:

Making myself smaller keeps the relationship peaceful.

Updated model:

Relationships that require self-erasure are unstable by design.

That distinction restored clarity.

Rebuild Internal Decision-Making

I started asking:

What do I actually feel?
What do I genuinely want?
What would I choose without fear of disapproval?
What response protects my integrity instead of preserving access?

The goal became internal alignment, not external management.

Allow Relationships to React to the Real Version of You

This part changes everything.

When authenticity returns:

some dynamics improve
some people adapt
some relationships collapse

But collapse is sometimes evidence that the relationship depended on distortion to survive.

What Changed

After identifying identity drift patterns:

self-awareness increased
emotional exhaustion decreased
boundaries became clearer
authenticity felt safer
internal consistency strengthened

And most importantly:

I stopped confusing self-suppression with emotional maturity.

Reframing Connection

Old model:

Keeping the relationship stable is proof of emotional success.

Updated model:

If maintaining connection requires abandoning myself, the stability is artificial.

That distinction restored self-respect.

Takeaway

Identity drift happens when preserving connection slowly becomes more important than preserving yourself.

The process is gradual enough to feel invisible until emotional exhaustion, resentment, or disorientation finally expose it.

Healing is not only about finding better relationships.

It is about refusing to disappear inside them.

Because the strongest connections are not built through self-erasure.

They are built through mutual reality.

Status
Identity drift detection: active
Self-abandonment patterns: identified
Internal alignment systems: restoring

Emotional Rollbacks: Why We Revert to Old Versions of Ourselves Around Certain People

Khali Sollis — Fri, 28 Aug 2026 12:00:00 +0000

How specific relationships reactivate archived behavioral patterns, why personal growth collapses in familiar emotional environments, and what happens when old dynamics override newer versions of yourself

One of the strangest parts of growth is realizing that progress is not always permanent.

You can spend months:

building boundaries
strengthening self-respect
regulating emotions
developing clarity
becoming more grounded

And then one interaction with the wrong person suddenly pulls an older version of you back online.

The version that:

overexplains
seeks validation
tolerates disrespect
abandons boundaries
becomes emotionally reactive

It feels immediate.

Automatic.

Almost involuntary.

That’s when I realized:

Some relationships function like emotional rollback systems.

The Bug: Emotional Rollbacks

In software systems, a rollback restores an application to a previous state after instability or incompatibility.

Human relationships can trigger similar regressions.

Certain people reactivate:

old coping mechanisms
outdated identities
emotional survival patterns
previous nervous system conditioning

You are no longer operating from your current architecture.

You are temporarily running archived behavioral code.

Why Growth Feels Fragile Around Certain People

Because personal growth is highly environment-dependent.

Growth developed in stable environments can destabilize inside familiar emotional systems.

Especially when those systems originally shaped your survival behaviors.

The nervous system remembers:

who you had to become
what kept you emotionally safe
what earned approval
what prevented rejection

And around certain people, those old patterns reactivate automatically.

Familiar Dynamics Override Conscious Intentions

This realization was uncomfortable:

Some people did not merely remind me of the past.

They reactivated it neurologically.

Suddenly:

my boundaries weakened
my clarity blurred
my emotional regulation decreased
my self-trust destabilized

Not because growth disappeared.

Because older emotional programming temporarily regained control.

The Identity Regression Loop

Certain relationships contain archived versions of you.

When re-entering those dynamics, the nervous system often retrieves:

the self associated with survival inside that relationship

Not:

the self you consciously became afterward

This is why emotionally charged environments can trigger disproportionate reactions.

The system is not responding only to the present moment.

It is responding to historical conditioning.

The Difference Between Love and Familiar Programming

This distinction changed everything:

Not every strong pull toward someone is emotional alignment.

Sometimes it is emotional familiarity.

And familiarity can feel intensely magnetic even when the dynamic is unhealthy.

The nervous system often interprets:

predictable emotional conditioning

as:

personal chemistry

Those are not always the same thing.

Old Roles Reappear Fast

Certain people unconsciously expect older versions of you to remain accessible.

The moment interaction resumes, the system attempts restoring previous roles:

the fixer
the caretaker
the appeaser
the overexplainer
the emotionally available version

If boundaries are not fully stabilized, the rollback happens quietly.

You start performing old identities without realizing it.

Emotional Muscle Memory

The body remembers relational patterns.

Especially repetitive ones.

Which means:

tone of voice
specific phrases
conflict dynamics
approval withdrawal
emotional inconsistency

can instantly reactivate stored nervous system responses.

Sometimes before conscious awareness even catches up.

Why Some People Feel “Hard to Resist”

Often because they are connected to unresolved emotional conditioning.

Not necessarily because they are uniquely compatible.

The attachment may actually be tied to:

validation wounds
approval patterns
abandonment fears
identity dependence
nervous system familiarity

The pull feels emotional.

But underneath it is often procedural.

The Cost of Emotional Rollbacks

The dangerous part is not temporary emotional discomfort.

It is identity destabilization.

Repeated rollbacks slowly erode:

self-trust
behavioral consistency
emotional stability
internal authority

You begin doubting the newer version of yourself because older patterns still activate under pressure.

The Realization

Eventually, I stopped asking:

Why do I become different around them?

And started asking:

What survival version of myself does this relationship reactivate?

That question exposed the actual architecture underneath the dynamic.

The Fix: Identity Stabilization

Healing required more than avoiding people.

It required stabilizing the newer version of myself strongly enough that older environments could no longer override it automatically.

Recognize the Rollback Early

The first signs were subtle:

overexplaining
anxiety spikes
compulsive responsiveness
self-abandonment
emotional hypervigilance

These behaviors became rollback indicators.

Not personality traits.

Separate Familiarity From Alignment

Old model:

This feels intense, therefore it must matter deeply.

Updated model:

Intensity can also indicate old nervous system programming.

That distinction prevented many regressions.

Protect the Updated Identity

Growth requires reinforcement.

The newer version of yourself must become:

practiced
repeated
defended
normalized

Otherwise stressful relationships will continuously reload archived behavior.

Stop Performing Old Roles for Familiar People

This part was critical.

Some relationships survive only if the old version of you remains accessible.

When the updated version appears:

boundaries
self-respect
emotional clarity
reduced availability

the dynamic often destabilizes.

That destabilization is information.

What Changed

After identifying emotional rollback patterns:

interactions became easier to evaluate objectively
emotional regressions shortened significantly
self-awareness increased
internal consistency improved

And most importantly:

I stopped mistaking regression for authenticity.

Reframing Growth

Old model:

If I revert around them, maybe this is who I really am.

Updated model:

Certain environments reactivate archived survival behaviors automatically.

That distinction restored self-trust.

Takeaway

Some people do not simply enter your life.

They preserve older versions of you emotionally, neurologically, and behaviorally.

And without awareness, those dynamics can quietly override years of growth.

Healing is not only about changing yourself in isolation.

It is about maintaining your updated identity inside environments that once trained you to abandon it.

Because growth becomes real when the old system no longer has permission to restore previous versions of you automatically.

Status
Emotional rollback detection: active
Archived survival behaviors: identified
Identity stabilization protocols: strengthening

Training the Return: What 111 Randomized Trials Reveal About Attention

Khali Sollis — Tue, 25 Aug 2026 14:00:00 +0000

You're twenty minutes into a function that finally makes sense. Not "I can explain it" sense — the deeper kind, where the whole call stack is loaded into working memory at once and you can feel the shape of the bug before you've located it. A message notification slides in. You answer it, because it takes ten seconds and you're a reasonable person. You come back to the editor.

The cursor is exactly where you left it. The code hasn't changed. But the thing that made it intelligible thirty seconds ago is gone, and you can feel yourself re-reading the same six lines without them landing. The interruption itself took ten seconds. Reconstructing the mental state it displaced takes considerably longer.

This is usually filed under "distraction" and treated as a discipline problem: turn off notifications, use a timer, try harder. But there's a separate, more interesting question buried in that experience, and it happens to be one that meditation researchers have spent the last decade quantifying: what happens when a cognitive system repeatedly practices noticing that attention has wandered, and returning it to a chosen target?

We Might Be Asking the Wrong Question About Focus

The common mental model of "good focus" is something like an uninterrupted beam: a mind that, once aimed, doesn't drift. Under that model, distraction is pure failure, and the goal of any attention-training practice would be to reduce how often the mind wanders in the first place.

Meditation research suggests a different, more testable idea worth taking seriously: mind-wandering may be something to expect rather than a defect to eliminate. What might actually be trainable isn't the absence of wandering. It's the loop around it: how reliably the system detects that it has drifted, disengages from the drift, and gets back to the intended target.

That reframe matters because it changes what a "good" trial looks like. If the goal is zero wandering, most people fail constantly. If the goal is a fast, low-friction return, then every wandering episode becomes a rep — and reps are something a nervous system can plausibly get better at, the way a muscle gets stronger through repeated, effortful contraction rather than through never being asked to contract.

To be clear about what kind of claim this is: no dataset in the meditation literature directly measures "return efficiency" as a stopwatch-timed variable and tracks it across training. This is an interpretive frame suggested by the shape of what these studies do measure — not a demonstrated mechanism. Keep that distinction in mind, because the rest of the article leans on it.

The Loop, as a Framework

Stripped down, the practice that most attention-focused meditation research examines looks like a simple repeating structure:

Choose a target → attention wanders → notice the wandering → disengage from it → return to the target → repeat.

Borrowing lightly from systems language: this isn't a bug being patched. It's closer to a control loop — a process that continuously checks whether the current state has diverged from an intended state, and if so, executes a correction. The "bug" framing implies something is broken. The loop framing implies something is running as designed, and the skill lives in how efficiently the correction step executes.

This is a useful lens, not a proven architecture. Neuroscience hasn't established that the brain literally implements this as a discrete, three-step algorithm. What it has established — carefully, and with real caveats — is what happens behaviorally and functionally when people practice something structurally similar to this loop, over and over, for weeks at a time. That's where the evidence gets specific.

What 111 Randomized Trials Actually Found

The most comprehensive test of this is a 2024 meta-analysis by Nur Hani Zainal and Michelle Newman, published in Health Psychology Review, which pooled 111 randomized controlled trials — 9,538 participants across 22 countries — comparing mindfulness-based interventions (MBIs) against no-treatment, waitlist, or active control conditions [1]. Unlike earlier reviews that mixed randomized and non-randomized designs, this one restricted itself to RCTs specifically to control for the confound of who chooses to meditate in the first place.

The analysis tested effects across fifteen distinct cognitive subdomains rather than treating "cognition" as one variable, which is what makes it useful here. The results split cleanly into two piles.

Domains that showed a positive effect: global cognition, executive attention, working memory accuracy, inhibition accuracy, cognitive-flexibility (shifting) accuracy, sustained attention accuracy, a measure of moment-to-moment consistency in sustained attention (intra-individual variability), and subjective cognitive functioning — i.e., how capable people felt their attention was [1]. Effect sizes ranged from small to large, with the strongest effects on global cognition — a composite score aggregating performance across the individual task domains measured, not a proxy for general intelligence or IQ.

Those improved domains — executive attention, working memory accuracy, inhibition, shifting, sustained attention — are what the trials actually measured: task-based accuracy scores on standardized cognitive tests. That's consistent with the return-loop interpretation developed above — goal-maintenance and course-correction are plausible ingredients in tasks that reward accurate inhibition and shifting — but it's worth being precise about the direction of the inference. The meta-analysis did not measure "noticing that attention has wandered" or "returning to a target" directly. It measured performance on lab cognitive tasks. The loop is a frame for interpreting why those particular tasks might improve, not a variable anyone tracked.

Domains that did not show a reliable effect: orienting, processing speed, verbal fluency, episodic memory, cognitive error rates, and the latency — as opposed to accuracy — components of working memory, inhibition, and shifting [1]. In plain terms: people didn't reliably get faster on these tasks. Accuracy moved on a specific cluster of goal-maintenance-flavored measures; raw processing speed and unrelated memory systems like recalling past events did not.

A few moderators are worth flagging, because they cut against a simple "just meditate more" reading. Effect sizes did not depend on how many sessions people did, how long the program ran, or completion rate [1]. They did depend on delivery format: instructor-led, face-to-face training outperformed self-guided formats, and studies of lower methodological quality tended to report larger effects than higher-quality ones [1] — a pattern meta-scientists treat as a caution flag, not a footnote. Effects were also somewhat weaker in older adults and in women, an asymmetry the authors note but don't fully explain.

None of this says mindfulness training makes people smarter, faster, or more broadly capable. It says something narrower and, honestly, more interesting: across a large, randomized evidence base, a specific cluster of goal-maintenance and course-correction functions improved, while several unrelated cognitive systems didn't move at all.

Training Isn't Transfer

If you've spent time optimizing anything — a build pipeline, a training loop, a query — the next idea will feel familiar: improving performance on the exact thing you practiced doesn't guarantee improvement on adjacent things, even things that feel related.

That's essentially what the null results above describe. Practicing "notice the mind has wandered, return to the breath" moved accuracy on tasks involving goal maintenance and course correction. It did not move verbal fluency or episodic memory, which draw on different systems and weren't obviously exercised by the practice. It's worth thinking of this the way you'd think about training specificity anywhere else: getting better at one exercise doesn't automatically transfer to a different one that merely looks related. That's an analogy, not a citation to a separate body of transfer-of-training research — but the pattern in this particular dataset fits it: the gains cluster tightly around the kind of goal-tracking process the practice actually exercises, and stop right at the edge of that cluster.

For anyone who's watched a model overfit to its training distribution, or watched a team get extremely good at one CI pipeline and then struggle the moment the stack changes, this shouldn't be surprising. But it's worth stating explicitly, because the popular version of meditation coverage almost never states it. And that gap — between what a specific, bounded finding supports and what a headline claims — is worth looking at directly.

Where the Abstraction Leaks

Here's a useful historical marker. In 2014, a landmark JAMA Internal Medicine review by Madhav Goyal and colleagues pooled 47 RCTs (3,515 participants) and found moderate evidence that mindfulness programs produced small-to-moderate reductions in anxiety and depression — but found the evidence for effects on attention specifically insufficient or low-quality at the time, alongside similarly weak evidence for sleep, substance use, and eating behavior [2]. Ten years and a much larger, attention-specific evidence base later, Zainal and Newman's 111-trial analysis found consistent, moderate effects on several attention-adjacent cognitive functions [1].

That's not necessarily a contradiction, and it isn't safe to read it as "the earlier review simply missed an effect that was there all along." The two analyses asked somewhat different questions, used different inclusion criteria and outcome measures, and drew on evidence bases separated by a decade. What the contrast reliably shows is how much the empirical landscape changed: by 2024, researchers could aggregate a far larger body of randomized evidence focused specifically on cognitive outcomes than existed in 2013. Whether that's the whole explanation for the different conclusions, or whether differences in methodology also played a role, isn't something these two papers alone can settle. It's also worth noting, per a 2022 review of 44 separate meta-analyses covering 336 RCTs, that effect sizes across the mindfulness literature broadly depend heavily on what the intervention is compared against — effects tend to look larger against passive controls such as waitlists than against credible active comparators (like structured exercise) [3]. That's a caveat for the whole field, not just cognition research, and it's a reason to hold effect sizes loosely rather than as fixed constants.

Here's where the leak happens, concretely. A careful sentence looks like: "Across 111 randomized trials, mindfulness-based interventions produced small-to-moderate improvements in executive attention, working memory accuracy, and sustained attention, without corresponding improvements in processing speed or episodic memory." That sentence is boring, bounded, and defensible.

The version that circulates instead is usually something closer to: "Science proves meditation makes you sharper." Every qualifier — which domains, what kind of intervention, delivered how, compared to what — evaporates in the retelling. The retracted-and-corrected literature on this topic offers a sharper cautionary example: a 2023 anatomical meta-analysis claiming mindfulness produces structural brain changes was retracted in 2025 after it emerged that four null-finding studies, representing 40% of the pooled participants, had been excluded from the analysis. The retraction notice states that this exclusion meant the analysis "was not designed to test the stated hypothesis" it claimed to address, and therefore could not support the paper's conclusions [4]. Forty percent of the participant pool disappearing from the analysis is, on its own, the whole story here. That paper is not cited elsewhere in this piece as evidence of anything, except as an illustration of how easily a headline-friendly brain claim can outrun what the underlying data supports.

What the Evidence Actually Permits Us to Say

Put together, a fair reading of this literature supports something like this: across randomized comparisons, mindfulness-based interventions were associated with improvements in accuracy on a specific cluster of goal-maintenance and course-correction-flavored cognitive tasks — executive attention, working memory, inhibition, shifting, sustained attention — without a matching association for processing speed, verbal or episodic memory, or anything resembling general intelligence. Reported effects appear to depend more on delivery quality than on dose, and — since lower-quality studies in this literature tended to report larger effects — the honest posture toward the size of any individual number here is a cautious one.

That's a smaller claim than "meditation makes you focused." It's also a more useful one, because it suggests a different way to think about the target of training: perhaps not the absence of distraction, but the fidelity of the return.

Whether that generalizes to a debugging session, a code review, or a deep-work block is an open, unanswered question — the trials measured laboratory cognitive tasks, not IDE behavior, and nothing here licenses a claim about programming performance specifically. What the data does support is a narrower shift in how to think about attention itself: not as a beam that either holds or breaks, but as a system that drifts by default — one where a cluster of accuracy-based, goal-maintenance-flavored functions responded to training, in a pattern consistent with, though not direct proof of, the idea that how well it comes back matters as much as whether it wanders at all.

It's also worth saying plainly, since the picture shouldn't be painted as uniformly benign: a 2020 systematic review of adverse events across meditation research found that among the subset of studies that specifically investigated harms, a majority reported at least one adverse event, most commonly anxiety, depressive symptoms, or unusual cognitive experiences [5]. That's not a reason for alarm about a training loop metaphor — it's a reminder that "attention training" in this literature refers to a real, sometimes intense psychological practice, not an abstraction, and the evidence should be read with that in mind.

References

[1] Zainal, N. H., & Newman, M. G. (2024). Mindfulness enhances cognitive functioning: A meta-analysis of 111 randomized controlled trials. Health Psychology Review, 18(2), 369–395. https://doi.org/10.1080/17437199.2023.2248222

[2] Goyal, M., Singh, S., Sibinga, E. M. S., Gould, N. F., Rowland-Seymour, A., Sharma, R., Berger, Z., Sleicher, D., Maron, D. D., Shihab, H. M., Ranasinghe, P. D., Linn, S., Saha, S., Bass, E. B., & Haythornthwaite, J. A. (2014). Meditation programs for psychological stress and well-being: A systematic review and meta-analysis. JAMA Internal Medicine, 174(3), 357–368. https://doi.org/10.1001/jamainternmed.2013.13018

[3] Goldberg, S. B., Riordan, K. M., Sun, S., & Davidson, R. J. (2022). The empirical status of mindfulness-based interventions: A systematic review of 44 meta-analyses of randomized controlled trials. Perspectives on Psychological Science, 17(1), 108–130. https://doi.org/10.1177/1745691620968771

[4] Siew, S., & Yu, J. (2025). Retraction note: Mindfulness-based randomized controlled trials led to brain structural changes: An anatomical likelihood meta-analysis. Scientific Reports, 15, 25545. https://doi.org/10.1038/s41598-025-11069-9

[5] Farias, M., Maraldi, E., Wallenkampf, K. C., & Lucchetti, G. (2020). Adverse events in meditation practices and meditation-based therapies: A systematic review. Acta Psychiatrica Scandinavica, 142(5), 374–393. https://doi.org/10.1111/acps.13225

Emotional Memory Leaks: Why Old Relationships Keep Reappearing During Vulnerable Moments

Khali Sollis — Fri, 21 Aug 2026 12:00:00 +0000

How unresolved emotional data resurfaces under stress, why loneliness reactivates archived attachments, and what happens when the mind mistakes familiarity for healing

I noticed something strange during difficult periods of my life.

The moment I became:

emotionally exhausted
lonely
overwhelmed
uncertain
disconnected from myself

old relationships started resurfacing in my mind.

Not randomly.

Predictably.

People I had not thought about in months suddenly reappeared internally:

memories
fantasies
conversations
nostalgia
imagined reconnections

At first, I thought it meant I still loved them.

Eventually, I realized something else was happening entirely.

The system was leaking emotional memory under stress.

The Bug: Emotional Memory Leaks

In software systems, memory leaks occur when unused or outdated data continues occupying system resources instead of being properly released.

Under pressure, inefficient systems begin resurfacing residual processes.

Human attachment behaves similarly.

Old emotional connections often reactivate during:

stress
loneliness
burnout
identity instability
emotional deprivation

Not necessarily because the relationship is still aligned.

Because the nervous system associates familiar attachment with temporary emotional relief.

Why Vulnerability Reactivates Old Attachments

The mind seeks familiarity during instability.

Especially emotional familiarity.

When life feels uncertain, the nervous system instinctively searches archived emotional memory for:

safety
validation
certainty
comfort
identity reinforcement

Even if the original relationship was unhealthy.

Familiarity can feel emotionally safer than uncertainty.

Nostalgia as Emotional Compression

This realization changed everything:

The mind does not store relationships objectively.

It compresses them emotionally.

During vulnerable moments, the brain often retrieves:

comforting fragments
emotionally intense highlights
symbolic moments of closeness

while minimizing:

inconsistency
incompatibility
emotional damage
unresolved pain

The archived memory becomes emotionally edited.

Why the Past Suddenly Looks Better During Hard Times

Because vulnerability distorts emotional evaluation.

The system starts comparing:

current discomfort

against:

selective emotional memory

Not against full reality.

The brain is not necessarily searching for truth.

It is searching for relief.

Emotional Regression Loops

Under stress, people often regress toward:

familiar emotional patterns
old attachment dynamics
previous coping mechanisms

Which means emotional vulnerability can reactivate dormant attachment systems.

The loop often looks like this:

stress
→ emotional depletion
→ nostalgic memory activation
→ fantasy reconstruction
→ emotional attachment resurgence

The attachment feels meaningful again.

But frequently, the timing—not the compatibility—is driving the intensity.

The Difference Between Missing Someone and Missing Regulation

This distinction became critical.

Sometimes I did not actually miss the person.

I missed:

emotional certainty
familiarity
distraction from pain
nervous system regulation
the temporary relief the attachment once provided

The mind confused:

emotional soothing

with:

relationship alignment

Those are not always the same thing.

Vulnerability Lowers Emotional Standards

This was difficult to admit.

During emotionally strong periods, certain relationships looked obviously misaligned.

During vulnerable periods, the same relationships suddenly appeared:

meaningful
rare
emotionally magnetic
impossible to replace

Not because the facts changed.

Because emotional depletion altered perception.

The Nervous System and Familiar Pain

The nervous system prefers predictable pain over unfamiliar uncertainty.

Which explains why people sometimes revisit:

emotionally unavailable partners
inconsistent relationships
destabilizing attachments

during difficult periods.

The system chooses:

known emotional territory

over:

unfamiliar emotional rebuilding

Even when the old territory caused suffering.

Why Healing Is Not Linear

One important realization:

Thinking about someone again does not automatically mean healing failed.

Sometimes vulnerability simply reactivates archived emotional pathways.

The appearance of memory is not necessarily evidence of destiny.

Nor proof that the relationship should resume.

Often, it is evidence that the nervous system is searching for familiar regulation during instability.

The Realization

Eventually, I stopped asking:

Why am I thinking about them again?

And started asking:

What emotional condition inside me triggered this memory resurgence?

That question changed the entire interpretation.

The Fix: Emotional Leak Management

Detachment became less about suppressing memories and more about understanding activation conditions.

Not every resurfacing emotion required action.

Some required observation.

Identify the Trigger State

I began noticing patterns.

Old attachments resurfaced most during:

exhaustion
loneliness
rejection
uncertainty
emotional dysregulation

The memory was often attached to vulnerability, not current compatibility.

Stop Romanticizing Emotional Regression

Old model:

If they came back into my mind, it must mean something profound.

Updated model:

Stress can reactivate familiar emotional pathways automatically.

That distinction prevented many unnecessary spirals.

Separate Emotional Memory From Present Reality

The memory might feel emotionally vivid.

That does not mean:

the relationship improved
the incompatibilities disappeared
the attachment became healthy

Memory intensity is not evidence of alignment.

Strengthen Present-Day Stability

The strongest protection against emotional memory leaks was rebuilding:

routine
identity
nervous system regulation
purpose
self-trust

Stable systems revisit old dependencies less frequently.

What Changed

After understanding emotional memory leaks:

nostalgia lost some of its authority
emotional relapses became easier to interpret
vulnerability stopped automatically triggering reconnection fantasies
self-awareness increased significantly

And most importantly:

I stopped mistaking emotional resurfacing for emotional truth.

Reframing Nostalgia

Old model:

If I miss them during hard moments, they must still belong in my life.

Updated model:

Emotional vulnerability often reactivates familiar attachment memories regardless of compatibility.

That distinction restored clarity.

Takeaway

Old relationships often resurface during vulnerable moments not because they are necessarily right for us, but because the nervous system associates them with familiarity, regulation, and temporary emotional relief.

The mind leaks archived attachment data under stress.

Especially when identity, stability, or emotional grounding weaken.

Healing is not the absence of memories.

It is learning not to confuse emotional resurfacing with emotional direction.

Because not every memory is a message.

Sometimes it is simply residual emotional code reappearing under pressure.

Status
Emotional memory leak detection: active
Nostalgia distortion awareness: increasing
Internal stability systems: strengthening

Phantom Notifications: Why Your Nervous System Still Waits for People Who Left

Khali Sollis — Fri, 14 Aug 2026 12:00:00 +0000

How emotional conditioning trains the body to anticipate unavailable people, why silence can feel physically destabilizing, and what happens when attachment rewires your internal alert system

Long after certain relationships ended, something strange kept happening.

I would feel my nervous system react before anything even occurred.

A phantom impulse.

A sudden anticipation.

An invisible expectation that:

they would text
they would appear
they would return
they would explain
they would finally become consistent

Sometimes my phone hadn’t even vibrated.

But internally, the system had already reacted.

That’s when I realized:

Attachment had trained my nervous system into anticipatory surveillance.

The Bug: Phantom Notifications

In technology systems, phantom notifications occur when the user perceives alerts that never actually happened.

The expectation becomes so conditioned that the brain generates the sensation automatically.

Emotionally, the same thing happens after unstable attachments.

The nervous system becomes trained to anticipate:

contact
validation
reassurance
emotional resolution

Even after the person is gone.

The Nervous System Learns People

One difficult truth:

The body memorizes emotional patterns faster than the conscious mind does.

Repeated attachment cycles create conditioned responses:

message received
→ dopamine spike
silence
→ anxiety activation
attention withdrawal
→ emotional destabilization

Over time, the nervous system stops responding to reality.

It starts responding to expectation.

Why Silence Feels So Loud

When emotional conditioning becomes strong enough, absence itself becomes stimulating.

Not peaceful.

Activating.

The mind starts monitoring:

unread messages
online activity
timing patterns
subtle changes in tone
possible signs of return

The attachment converts attention into vigilance.

Intermittent Reinforcement Creates Obsession

Consistent people calm the nervous system.

Inconsistent people often hijack it.

Because unpredictability intensifies emotional conditioning.

The cycle looks like this:

distance
→ anxiety
unexpected attention
→ relief
relief
→ stronger attachment

The nervous system becomes addicted to emotional resolution.

Not stability.

Emotional Gambling Systems

This realization was uncomfortable:

Some attachments operate like variable reward systems.

You never know:

when attention will arrive
when affection will return
when emotional warmth will appear again

That unpredictability creates compulsive monitoring behavior.

The brain starts chasing:

possibility

instead of evaluating:

consistency
Phantom Hope

Eventually, the nervous system begins generating anticipatory responses automatically.

You:

check the phone unnecessarily
reread old conversations
imagine future contact
mentally rehearse reconciliation scenarios

Not because anything happened.

Because the body became conditioned to expect emotional stimulation from that source.

The Attention Hijack

The most damaging part wasn’t the relationship itself.

It was the continuous occupation of attention.

Part of my mind remained permanently allocated toward:

waiting
monitoring
anticipating
interpreting silence

Even during:

work
rest
conversations
ordinary moments

The attachment had become background processing.

When Waiting Becomes Identity

At some point, I realized I had unconsciously organized parts of my emotional life around anticipation.

Waiting became normalized.

Waiting for:

clarity
consistency
reassurance
effort
emotional reciprocation

The nervous system adapted to uncertainty so completely that peace began feeling unfamiliar.

The Difference Between Connection and Conditioning

This distinction changed everything:

Not every strong emotional reaction is evidence of deep compatibility.

Sometimes it is evidence of strong conditioning.

intensity
≠
alignment
obsession
≠
love
anticipation
≠
security

The nervous system can become highly activated around emotionally inconsistent people.

That activation is not always meaningful.

Sometimes it is simply conditioned survival behavior.

The Realization

Eventually, I stopped asking:

Why can’t I stop thinking about them?

And started asking:

Why has my nervous system mistaken unpredictability for emotional importance?

That question exposed the architecture underneath the attachment.

The Fix: Nervous System Deconditioning

Detachment became less about suppressing emotion and more about retraining internal responses.

Not instantly.

Repeatedly.

Stop Rewarding Monitoring Behavior

Every time I:

checked their profile
reread messages
searched for signs
mentally revisited the connection

the nervous system interpreted it as:

this attachment still matters for survival

Reducing reinforcement weakened the loop.

Normalize Calm

This part was surprisingly difficult.

Chaos had become emotionally familiar.

Peace initially felt:

boring
emotionally flat
suspiciously quiet

The nervous system had adapted to volatility.

Stability required recalibration.

Interrupt Automatic Anticipation

Whenever phantom expectation appeared:

Maybe they’ll...

I stopped feeding the narrative.

Not with suppression.

With observation.

This is conditioned anticipation.
Not evidence.

That distinction reduced emotional spirals significantly.

Reinvest Attention Into Present Reality

Attention is finite infrastructure.

Every hour spent emotionally monitoring unavailable people is attention removed from:

health
growth
creativity
self-respect
future construction

The nervous system stabilizes when attention returns to reality instead of emotional prediction loops.

What Changed

After interrupting phantom notification conditioning:

emotional vigilance decreased
compulsive checking weakened
nervous system stability improved
attention fragmentation reduced

And most importantly:

Silence stopped feeling like emotional danger.

Reframing Attachment

Old model:

If I still react strongly, the connection must still matter.

Updated model:

Strong reactions can also indicate conditioned emotional surveillance.

That distinction restored clarity.

Takeaway

Some relationships continue affecting people long after they end because the nervous system was conditioned to anticipate emotional resolution from inconsistent sources.

The body keeps waiting.

The mind keeps scanning.

The attention system stays partially trapped in expectation.

Healing begins when the nervous system no longer interprets someone’s absence as an emergency requiring continuous monitoring.

Because peace is impossible while your internal alert system is still waiting for notifications from people who already left.

Status
Phantom notification loops: weakening
Emotional surveillance behavior: interrupted
Nervous system recalibration: active

Ghost Dependencies: The Relationships That Still Control You After They End

Khali Sollis — Fri, 07 Aug 2026 12:00:00 +0000

How unresolved emotional architecture continues operating after a relationship is gone, why absence does not equal detachment, and what happens when invisible dependencies keep running in the background

I used to think relationships ended when communication stopped.

That turned out to be false.

Some relationships continue operating long after:

the conversations end
the access disappears
the contact stops
the person leaves

Not externally.

Internally.

The visible connection may be gone.

But the dependency architecture remains active.

The Bug: Ghost Dependencies

In software systems, ghost dependencies are outdated or indirect dependencies that continue influencing the system even after the primary component is supposedly removed.

The application appears independent.

But hidden processes still rely on old structures.

Human attachment behaves similarly.

A relationship can end while invisible emotional dependencies continue running:

validation dependency
identity dependency
emotional regulation dependency
routine dependency
fantasy dependency

The person leaves.

The dependency survives.

Why “Moving On” Sometimes Fails

Because physical absence does not automatically remove emotional infrastructure.

You may stop talking to someone while still:

mentally consulting them
emotionally orienting around them
seeking their validation internally
imagining their reactions
structuring your worth around their perception

The relationship stops externally.

The system remains connected internally.

Hidden Emotional Processes

This realization was uncomfortable:

Some people were no longer in my life, but they were still influencing:

my confidence
my decisions
my emotional baseline
my self-worth
my nervous system responses

The dependency had become ambient.

Invisible.

Automatic.

The Lingering Architecture of Attachment

Relationships often install unconscious behavioral systems.

For example:

their approval
→ emotional stability
their distance
→ anxiety response
their attention
→ self-worth increase

Over time, the nervous system automates these associations.

Even after the relationship ends, the responses remain operational.

Emotional Residue

The difficult part is that ghost dependencies rarely feel obvious.

They show up subtly:

checking whether they would approve
mentally explaining yourself to them
comparing new people to them
replaying unresolved scenarios
feeling emotionally destabilized by reminders

The person becomes psychologically embedded beyond active contact.

When Someone Becomes Part of Your Internal Operating System

Some attachments persist because the relationship stopped being relational.

It became structural.

Their presence shaped:

identity
routines
emotional regulation
future planning
self-perception

When they leave, the system experiences destabilization because too many internal functions were outsourced externally.

The Illusion of Closure

Many people believe closure comes from:

final conversations
explanations
apologies
mutual understanding

Sometimes closure is much more technical than emotional.

Closure occurs when the dependency stops running.

Not when the story feels perfectly resolved.

The Nervous System Problem

This part matters more than most people realize.

The nervous system memorizes emotional patterns:

anticipation
reassurance cycles
conflict rhythms
emotional highs and lows
attachment reinforcement

When the relationship disappears, the body continues expecting those cycles.

Which is why absence can initially intensify obsession instead of reducing it.

The system is still searching for the missing dependency.

Ghost Dependencies and Identity

One of the strongest hidden dependencies was identity-based.

Without realizing it, I had attached parts of myself to:

being chosen
being needed
being understood
being emotionally significant to someone

When the relationship ended, those identity anchors destabilized too.

Part of the pain wasn’t losing them.

It was losing the version of myself that existed around them.

The Recursive Replay Loop

Ghost dependencies survive through repetition.

The system continuously reloads:

memory
→ emotional activation
→ mental replay
→ reinforcement

The loop keeps the dependency alive despite external absence.

Which means time alone does not necessarily heal attachment.

Uninterrupted reinforcement loops prolong it.

The Realization

Eventually, I stopped asking:

Why am I still affected by this?

And started asking:

What emotional functions am I still outsourcing to someone who is no longer here?

That question exposed the hidden architecture underneath the attachment.

The Fix: Dependency Removal

Detachment became less emotional and more structural.

The goal was not suppressing memories.

The goal was removing unauthorized control pathways.

Identify the Emotional Function

I stopped viewing the attachment as one giant emotional mass.

Instead, I broke it down technically:

validation
comfort
identity reinforcement
certainty
distraction from loneliness
emotional stimulation

Clarity reduced emotional fog immediately.

Stop Consulting Absent People Internally

This was subtle but critical.

I noticed how often I mentally referenced:

What would they think?

Even in their absence.

That meant internal authority had not fully returned to me.

Rebuild Internal Sources

Dependencies weaken when the functions become internally supported.

Instead of outsourcing:

worth
peace
direction
emotional regulation

the system slowly relearns self-generated stability.

Not perfectly.

But progressively.

Interrupt Emotional Replay Cycles

The nervous system strengthens what it repeatedly rehearses.

Constant replay keeps old architecture active.

Reducing:

memory spirals
fantasy loops
emotional revisiting
symbolic attachment rituals

gradually weakens dependency pathways.

What Changed

After identifying ghost dependencies:

emotional triggers became easier to understand
obsessive thinking decreased
identity stabilized
internal authority strengthened

And most importantly:

The relationship stopped unconsciously controlling my emotional environment.

Reframing Detachment

Old model:

I still care, therefore I am still connected.

Updated model:

Emotional residue is not proof that the dependency should remain active.
Takeaway

Some relationships continue controlling people long after they end because the visible relationship disappeared while the hidden dependencies remained installed.

Healing is not only emotional.

It is architectural.

The goal is not becoming cold or forgetting the connection existed.

The goal is removing outdated emotional systems that still control your inner environment without your permission.

Because absence alone does not create freedom.

Dependency removal does.

Status
Ghost dependency detection: active
Unauthorized emotional pathways: identified
Internal authority restoration: in progress