Skip to content

Counterfactual Proximity Weighting

Prime #
757
Origin domain
Cognition Judgment Decision Making
Subdomain
reward and signal learning → Cognition Judgment Decision Making

Core Idea

Counterfactual proximity weighting is the structural pattern in which the internal signal an agent or system assigns to an outcome is not a function of the outcome alone but of the distance from that outcome to a nearby counterfactual outcome of different value. A near miss carries a near-reward signal even though no reward was delivered; a near-catastrophe carries a near-loss signal even though no harm occurred. The pattern concerns signal generation, not interpretation: the internal update — to behaviour, attention, a risk register, an investment posture — is graded by proximity to a counterfactual rather than by the outcome's realised value alone. When the actual outcome was a clean miss but a catastrophe lay close by in the relevant state space, the system updates as if the catastrophe partially happened.

The pattern carries a small set of structural commitments. There is an agent or system generating an internal signal in response to an outcome; a nearby counterfactual outcome of meaningfully different value; a proximity measure in the relevant state space — causal distance, perceptual similarity, temporal nearness; a decreasing function of that distance that weights the counterfactual's value into the signal; and a downstream effect calibrated to the signal rather than to the realised outcome alone. The structure admits a compact model: with r(o) the realised value of outcome o and c(o) the value of the nearest counterfactual at distance d, the internal signal is well-fit by s(o) = α·r(o) + (1 − α)·w(d)·c(o), where w decreases in distance. The classical assumption α = 1 — signal depends only on the realised outcome — is the limiting case; the prime is everything to its left. Critically, the signal is not irrational when outcomes are noisy: a near miss genuinely carries information about how close to a failure boundary the agent was operating. The pattern becomes maladaptive only when the proximity weight is miscalibrated relative to the true counterfactual probability, which is the substrate-specific question.

How would you explain it like I'm…

So Close It Counts

Imagine you almost catch a ball but it slips right through your fingers. You did not catch it, yet your heart jumps as if you nearly did, just because you were so close. A near miss feels almost like the real thing. How close you came changes how big the feeling is.

The Near-Miss Feeling

Counterfactual proximity weighting means the signal a system feels about an outcome depends not just on what happened, but on how close a different outcome was. A near miss carries a little of the reward you almost got; a near-disaster carries a little of the loss that almost happened, even though nothing bad actually occurred. The closer the other outcome was, the more strongly it colours the feeling. This is about how the signal is made, not how it is read later: your behaviour or worry updates as if part of the close-by outcome really happened. And it is not silly, because in a noisy world a near miss really does tell you that you were operating close to a failure line.

Distance to the Almost

Counterfactual proximity weighting is the pattern where the internal signal an agent assigns to an outcome is not a function of the outcome alone but of the distance from it to a nearby counterfactual outcome of different value. A near miss carries a near-reward signal though no reward was delivered; a near-catastrophe carries a near-loss signal though no harm occurred. It concerns signal generation, not interpretation: the internal update, to behaviour, attention, or a risk register, is graded by proximity to a counterfactual rather than by the realised value alone. When the actual outcome was a clean miss but a catastrophe lay close by in the relevant state space, the system updates as if the catastrophe partly happened. This is not irrational when outcomes are noisy, because closeness genuinely carries information about how near a failure boundary you were; it becomes maladaptive only when the proximity weight is miscalibrated relative to the true counterfactual probability.

 

Counterfactual proximity weighting is the structural pattern in which the internal signal an agent or system assigns to an outcome is not a function of the outcome alone but of the distance from that outcome to a nearby counterfactual outcome of different value. A near miss carries a near-reward signal even though no reward was delivered; a near-catastrophe carries a near-loss signal even though no harm occurred. The pattern concerns signal generation, not interpretation: the internal update, to behaviour, attention, a risk register, an investment posture, is graded by proximity to a counterfactual rather than by the realised value alone. The commitments are an agent generating an internal signal in response to an outcome, a nearby counterfactual of meaningfully different value, a proximity measure in the relevant state space (causal distance, perceptual similarity, temporal nearness), a decreasing function of that distance weighting the counterfactual's value into the signal, and a downstream effect calibrated to the signal rather than the realised outcome. A compact model fits it: with r(o) the realised value and c(o) the value of the nearest counterfactual at distance d, the signal is s(o) = alpha*r(o) + (1 - alpha)*w(d)*c(o), where w decreases in distance. The classical assumption alpha = 1 (signal depends only on the realised outcome) is the limiting case; the prime is everything to its left. The signal is not irrational under noise, since a near miss genuinely carries information about how close to a failure boundary the agent was operating; it becomes maladaptive only when the proximity weight is miscalibrated relative to the true counterfactual probability.

Structural Signature

an agent or system generating an internal signal — a realised outcome of some value — a nearby counterfactual outcome of meaningfully different value — a proximity measure in the relevant state space — a distance-decreasing weight folding the counterfactual into the signal — a downstream effect calibrated to the signal rather than the realised outcome

The pattern is present when each of the following holds:

  • A signal-generating agent. Something produces an internal update — to behaviour, attention, a risk register, an investment posture — in response to an outcome.
  • A realised outcome. An actual result occurs with its own value r(o); under the classical outcome-only view, only this would matter.
  • A nearby counterfactual. A different outcome of meaningfully different value c(o) — a jackpot just missed, a catastrophe just avoided — lies close by in the relevant state space.
  • A proximity measure. A distance metric exists in that space: causal-chain branching, perceptual similarity, score margin, temporal nearness.
  • A distance-decreasing weight. A function w(d) that falls with distance folds the counterfactual's value into the signal: s(o) = α·r(o) + (1 − α)·w(d)·c(o), with the outcome-only case recovered at α = 1.
  • A signal-calibrated downstream effect. The behavioural update tracks s, not r — the system updates as if the nearby counterfactual partially happened.

The components compose so that signal generation separates from outcome interpretation: a near miss carries genuine information about operating margin when outcomes are noisy, so the weighting is normatively appropriate up to the accuracy of w(d). The pathology is not the response but a single auditable parameter — whether the proximity weight tracks true counterfactual probability or mere perceptual vividness.

What It Is Not

  • Not counterfactual reasoning. counterfactual_reasoning is the cognitive operation of considering what would have happened under different conditions; counterfactual proximity weighting is the signal-generation mechanism in which an internal update is graded by distance to a nearby counterfactual of different value. Reasoning constructs the alternative; proximity weighting weights it into a reward/loss signal.
  • Not anchoring. anchoring is the pull of an initial reference value on a judgement; counterfactual proximity weighting grades a signal by how near an alternative outcome was, not by adjustment from a starting anchor.
  • Not attention. attention is the selective allocation of processing; counterfactual proximity weighting is about how an outcome's internal signal is computed (weighting a nearby counterfactual), though it may drive attention downstream.
  • Not conformity. conformity is alignment of behaviour to a group; counterfactual proximity weighting is an individual signal-generation pattern keyed to a counterfactual's proximity, with no social reference required.
  • Not regret as such. Regret and relief are instances of the pattern (signals scaling with counterfactual proximity), not the structure itself, which is substrate-general across reward, loss, and engineered systems.
  • Common misclassification. Reading a response disproportionate to the realised outcome as irrational. Catch it by asking what counterfactual lay nearby and how its value is weighted in; when outcomes are noisy a near miss carries genuine information, so the question is whether the proximity weight tracks true counterfactual probability or mere perceptual vividness.

Broad Use

  • Operant learning and gambling: the near-miss effect — a slot machine showing "two cherries and a lemon" produces approach behaviour and dopaminergic response resembling a partial win, even though nothing was paid; the graded counterfactual proximity drives re-engagement.
  • Decision theory and emotion: counterfactual regret (how much better could it have gone?) and relief (how much worse?) scale with the proximity of the alternative, not its mere existence; regret-driven choice models formalise the weighting.
  • Safety engineering: leading-indicator weighting of near-misses (the Heinrich pyramid, state-safety programmes) enters near-catastrophes into the risk register at a fraction of the catastrophe's cost, proportional to their proximity in the causal chain.
  • Sports and competition: almost-victory drives re-engagement more than clear defeat; the near-counterfactual win carries a partial-reward signal that powers subsequent practice and investment.
  • Negotiation: an offer "just barely" rejected anchors the next round more than one wildly off-base; the narrow gap to acceptance functions as a partial-acceptance signal.
  • Driving and piloting: a near-collision recalibrates risk perception disproportionate to the objective hazard density, because the proximity-weighted near-loss signal drives attention recalibration.

Clarity

The label distinguishes the generation of an internal signal from its interpretation. The naive lens treats outcomes as binary — won or lost, harmed or unharmed — and reads an agent's subsequent behaviour as a response to that outcome; behaviour that does not match the realised outcome then looks irrational. The proximity-weighted lens reveals an entire layer of pre-outcome signal: how close was the counterfactual? This is what makes near-miss data informative about risk and what makes near-victory motivating about engagement, and it reframes a class of apparently irrational responses as correct responses to a counterfactual-proximity signal the outcome record does not capture. The clarifying force is to separate the realised value of an outcome from the proximity-weighted signal the system actually acts on, and to make the proximity weight itself the auditable object — is it calibrated to the true counterfactual probability, or to perceptual vividness? Once named, the gambler chasing near-misses, the over-corrector driven by regret, and the post-near-miss safety hyper-vigilance are visibly the same mechanism rather than separate quirks.

Manages Complexity

The pattern reduces a confused mix of "irrational" behaviours to a single mechanism — the system is responding to a counterfactual-proximity signal the outcome record does not contain — and compresses the intervention space into three families keyed to the direction of the desired change. To dampen the response, suppress the salience of the counterfactual neighbour: regulators removing near-miss displays from gaming machines, debriefings that focus on the actual outcome rather than the avoided one, techniques for interrupting regret rumination. To amplify the response, surface the counterfactual neighbour: safety programmes that publicise close-call statistics, leadership debriefs that walk through "what could have happened." To weight it correctly, ground the proximity weight in an explicit causal model rather than perceptual proximity: fault-tree analysis weights near-misses by failure-mode branching, not by emotional vividness. The compression is that a single parameter — the proximity weight and how it is set — captures near-miss reinforcement, regret and relief, leading-indicator safety weighting, and motivational asymmetry, so an intervention designer reasons about that one lever rather than about a catalogue of unrelated phenomena. The same move (suppress or surface the counterfactual neighbour) reliably shifts the response across substrates.

Abstract Reasoning

The prime trains a reasoner to ask, of any update that seems disproportionate to the realised outcome: what counterfactual lay nearby, how close was it, and how is its value being weighted into the internal signal? The compact model s(o) = α·r(o) + (1 − α)·w(d)·c(o) makes the reasoning explicit — the behaviour is driven by s, not by r, and the classical outcome-only view is recovered only at α = 1. The non-obvious move the prime licenses is to treat a "near miss" as carrying genuine information rather than as a non-event: when outcomes are noisy, proximity to a failure boundary is real evidence about operating margin, so the proximity-weighted signal is normatively appropriate up to the accuracy of its weight. This reframes the diagnostic question from "is this response rational?" to "is the proximity weight calibrated to the true counterfactual probability, or to perceptual vividness?" — relocating the pathology from the response itself to a single, auditable parameter. The reasoning generalises across reward and loss, across stochastic and non-stochastic counterfactuals, and across agents and engineered systems, because all that the structure requires is an internal signal, a nearby counterfactual of different value, and a distance metric that weights one into the other.

Knowledge Transfer

The structure ports across substrates that share nothing else. A gambling regulator who understands the prime predicts that reducing near-miss displays will dampen play, and predicts the same effect across every reward-uncertainty substrate — loot boxes, retention loops, prediction-error-driven engagement systems — because the mechanism, not the surface, is what transfers. A safety officer who understands it extracts one intervention catalogue across aviation, medicine, nuclear, and industrial process safety: surface close-call data, weight it by causal proximity rather than vividness, and audit organisational responses to the graded signal. A negotiator who understands it can both predict and shape the anchoring effect of a narrowly rejected offer on the next round. The role mappings transfer directly — realised outcome ↔ no payout / actual loss / non-event / clear defeat; nearby counterfactual ↔ the jackpot just missed / the catastrophe just avoided / the win just lost; proximity measure ↔ symbol distance / causal-chain branching / score margin / temporal nearness; internal signal ↔ dopaminergic response / regret intensity / risk-register weight / motivation. The intervention vocabulary — suppress or surface the counterfactual neighbour, ground the weight in an explicit causal model — is invariant; the only substrate-specific work is identifying the relevant state space and the true counterfactual probability against which the weight should be calibrated. The transferred and non-obvious prediction is that the same intervention, suppressing counterfactual salience, reliably reduces the response wherever the structure holds — which is the prime's strongest cross-substrate evidence, since a gaming regulator, a safety officer, and a clinician arrive at the identical lever from unrelated literatures. The complement is equally portable: where the response is too weak, surfacing the counterfactual neighbour amplifies it, and where it is merely miscalibrated, replacing perceptual proximity with a causal model corrects it without abolishing the informative signal.

Examples

Formal/abstract

The compact signal model the prime supplies can be worked as a decision-theoretic instance, making every role explicit and the calibration question precise. Let an agent face an outcome o whose realised value is r(o) — say a clean loss, r(o) = 0. A nearby counterfactual outcome of meaningfully different value exists: a large win, c(o), that lay close by in the relevant state space. The proximity measure d quantifies that closeness — how few state transitions, or how small a margin, separated the realised loss from the counterfactual win. A distance-decreasing weight w(d), large when the counterfactual was near and falling toward zero as it recedes, folds the counterfactual's value into the internal signal: s(o) = α·r(o) + (1 − α)·w(d)·c(o). The classical outcome-only view is the limiting case α = 1, under which the agent updates only on r(o) and a clean loss produces no reinforcement. The prime is everything to the left of that limit: with α < 1 and a near counterfactual (w(d) large), the signal s(o) is substantially positive even though nothing was won, and the downstream effect — the behavioural update, the propensity to re-engage — tracks s, not r. The decisive analytical point is that this is not irrational when outcomes are noisy: proximity to a high-value boundary is genuine evidence about the agent's operating margin, so weighting c(o) into the signal is normatively appropriate up to the accuracy of w(d). The pathology is therefore relocated from the response to a single auditable parameter: does w(d) track the true counterfactual probability (the actual chance the win would have occurred), or merely perceptual vividness (how salient the near-win looked)? When w(d) is anchored to vividness rather than probability, the same model that explains adaptive risk-sensitivity also explains maladaptive chasing — the difference is one calibration term, not two mechanisms.

Mapped back: The signal model instantiates every role of the signature — a signal-generating agent, a realised outcome r(o), a nearby counterfactual c(o), a proximity measure d, a distance-decreasing weight w(d) folding the counterfactual into s(o), and a downstream effect calibrated to s rather than r — and shows the prime's central separation of signal generation from outcome interpretation, with the pathology isolated to whether the weight tracks probability or vividness.

Applied/industry

Slot-machine near-miss design and aviation near-miss safety reporting are the same counterfactual-proximity-weighting object on a gambling-behaviour and a safety-engineering substrate, and they run the prime's lever in opposite directions, which is itself the strongest cross-substrate evidence. In the gambling case the agent is the player; the realised outcome is no payout (r = 0); the nearby counterfactual is the jackpot just missed; the proximity measure is symbol distance — "two cherries and a lemon" on the payline; and the internal signal is a dopaminergic, approach-driving response resembling a partial win. The machine is engineered to maximise w(d)'s perceptual vividness — displaying near-misses far more often than chance would produce — so the proximity weight is deliberately decoupled from the true (near-zero) counterfactual probability, driving re-engagement. The prime's diagnosis names this exactly: the response is not the player's irrationality but a proximity-weighted signal the outcome record does not contain, miscalibrated by design. The intervention follows from the suppress the counterfactual neighbour family — regulators removing or limiting near-miss displays reliably dampens play. In the aviation case the agent is the safety organisation; the realised outcome is a non-event (no crash); the nearby counterfactual is the catastrophe just avoided (a runway incursion that came within seconds of collision); the proximity measure is causal-chain branching — how few failures separated the non-event from disaster; and the internal signal is the weight that incident enters in the risk register. Here the desired move is the opposite — surface and amplify the counterfactual neighbour by publicising close-call data — and weight it correctly by grounding w(d) in fault-tree causal proximity rather than how frightening the event felt. A gambling regulator and a safety officer, arriving from unrelated literatures, reach the identical lever — manipulate the salience and calibration of the counterfactual neighbour — which is the prime's clearest demonstration that the mechanism, not the surface, is what transfers.

Mapped back: Near-miss slot design and aviation near-miss reporting are the same proximity-weighting mechanism as the formal model — an internal signal graded by distance to a nearby counterfactual of different value — run in opposite directions: gambling miscalibrates w(d) toward vividness to amplify engagement, while safety engineering re-grounds w(d) in causal probability to amplify a genuinely informative signal, both acting on the prime's single auditable parameter.

Structural Tensions

T1 — Proximity Weight as Calibrated versus Vivid (Measurement). The prime locates the entire pathology in one parameter: whether w(d) tracks true counterfactual probability or mere perceptual vividness. The two diverge sharply exactly where it matters — a near-miss that looked close but was causally remote, or one that felt routine but sat one failure from disaster. The failure mode is letting salience set the weight, amplifying vivid-but-improbable counterfactuals and ignoring drab-but-likely ones. Diagnostic: compare the proximity felt to the proximity a causal model assigns; where vividness and probability part company, the signal is miscalibrated, and grounding w(d) in an explicit causation model, not perception, is the corrective.

T2 — Counterfactual Distance versus the Right State Space (Scopal). Proximity is defined in "the relevant state space," but which space is relevant is itself contested — symbol distance, causal-chain branching, temporal nearness, and perceptual similarity can disagree about how near the counterfactual was. The failure mode is measuring proximity in a convenient but wrong metric (the slot's visual symbol distance, which has nothing to do with payout probability). Diagnostic: ask whether the distance metric used is the one that actually governs the counterfactual's likelihood; a near-miss can be close in one space and far in another, and choosing the space that flatters salience over the one that tracks probability is how the weight goes wrong before any calibration.

T3 — Informative Signal versus Manufactured Signal (Sign/Direction). The prime rightly defends near-miss signals as genuinely informative under noise — but the same mechanism is an attack surface: an adversary (a slot designer, a manipulator) can manufacture counterfactual proximity to inject a signal that carries no real information. The failure mode is trusting a proximity-weighted update that an interested party engineered. Diagnostic: ask whether the counterfactual's apparent nearness arose from the world or was constructed by someone who benefits from the response; where proximity is manufactured, the signal is adversarial noise wearing the form of evidence, and incentive analysis of who shaped the near-miss matters more than the signal's felt informativeness.

T4 — Single Counterfactual versus a Distribution (Scalar). The model weights the nearest counterfactual of different value, but real outcomes sit amid a whole distribution of counterfactuals at varying distances and values. Fixating on the single most salient neighbour ignores the many others that should also enter the signal. The failure mode is over-weighting one vivid near-catastrophe while a cluster of moderately-near, moderately-bad counterfactuals — collectively more important — goes unregistered. Diagnostic: ask whether the signal reflects the nearest counterfactual or an integral over the counterfactual distribution; where many alternatives lie nearby, a single-neighbour weighting mis-sizes the true risk, and the proper signal is an expectation, not a one-point read.

T5 — Adaptive Recalibration versus Persistent Distortion (Temporal). A near-miss appropriately recalibrates risk perception in the moment, but the weighting can persist long after the operating conditions that justified it have passed — a single near-collision driving years of disproportionate caution, regret rumination that will not decay. The failure mode is a one-time proximity signal that fails to fade, distorting behaviour indefinitely. Diagnostic: ask whether the elevated response decays as new (uneventful) outcomes accumulate or stays locked at the near-miss level; an adaptive signal updates downward with subsequent evidence, and a feedback loop that never relaxes the weight converts a useful recalibration into a permanent bias.

T6 — Signal Generation versus Downstream Over-Correction (Coupling). The prime carefully scopes itself to signal generation, treating the downstream behavioural effect as merely calibrated to the signal. But the coupling between signal and action can amplify a well-calibrated signal into a harmful over-correction — a correctly-weighted near-loss producing a swerve more dangerous than the avoided hazard, a justified regret driving a reckless make-up bet. The failure mode is a sound signal feeding an unsound response. Diagnostic: separate whether w(d) is calibrated (signal question) from whether the action it drives is proportionate (response question); even a perfectly weighted counterfactual signal needs a feedback-damped downstream policy, since the prime guarantees an informative signal, not a safe reaction to it.

Structural–Framed Character

Counterfactual Proximity Weighting sits at the hinge of the structural–framed spectrum — a genuine hybrid, with an aggregate of 0.5. There is a clean relational core: a signal graded by distance to a nearby counterfactual of different value, captured compactly as s(o) = α·r(o) + (1 − α)·w(d)·c(o), a model that is mathematically substrate-free. But the prime is about signal generation by an agent, and the diagnostics split between that structural model and its agent-bound, mildly affective framing.

The one criterion at full weight is decisive for the grade. Its human_practice_bound score is maximal (1.0): the pattern is inherently agent-side — every instance (near-miss gambling, regret and relief, leading-indicator safety weighting, near-victory motivation, near-collision recalibration) requires a system that generates an internal signal in response to an outcome, and there is no purely physical or biological substrate in which a "counterfactual of different value" is weighted into a "reward/loss signal" without an agent doing the weighting.

Three criteria carry partial weight, and one is zero, which together land the aggregate at 0.5. Its institutional_origin is 0.0: the mechanism is not minted by any institution — it is a cognitive/computational regularity. The vocabulary travels only halfway (0.5): "signal," "counterfactual," "proximity," "weight" port across reward learning, safety engineering, and negotiation, but a regret-and-reinforcement lexicon comes along. Its evaluative load is mixed (0.5): a mild affective framing rides with it — regret, relief, near-miss "chasing" — though the prime works to keep the core normatively neutral, insisting the signal is informative-not-irrational and locating any pathology in a single auditable calibration parameter. And invoking it is part recognition, part import (0.5): one can recognise a proximity-graded signal as a present mechanism, but naming it brings along the reward-signal interpretive frame. The honest reading is that the s(o) model is a real, substrate-free relational skeleton — which is why the prime sits no further toward framed than the middle — but it exists only in signal-generating agents and carries a mild affective frame, which is exactly the hybrid character the 0.5 aggregate records.

Substrate Independence

Counterfactual Proximity Weighting is a moderately substrate-independent prime — composite 3 / 5 on the substrate-independence scale. The structural core — grading an internal signal (reward, regret, alarm) by how close the actual outcome came to a salient counterfactual, so a near-miss registers far more strongly than a distant one — recurs in near-miss gambling (the almost-win that drives play), regret and relief, safety leading indicators (counting near-accidents), sports (the shot that just missed), negotiation (the deal that nearly closed), and driving behaviour. Its transfer evidence is real across these named instances. What holds every component to the middle is that the pattern lives on cognitive and behavioural substrates: it presupposes an agent with an internal valuation that can be modulated by imagined alternatives, so there is no clean physical or biological-mechanism instance in which the weighting operates with the same force. Domain breadth is bounded to behavioural settings; structural abstraction is held at moderate because the signature carries a cognitive valuation-and-counterfactual frame rather than a medium-neutral mechanism; and transfer evidence, while genuine, stays within the behavioural family. That cognitive-behavioural ceiling is what fixes all four sub-scores, and the composite, at a moderate 3.

  • Composite substrate independence — 3 / 5
  • Domain breadth — 3 / 5
  • Structural abstraction — 3 / 5
  • Transfer evidence — 3 / 5

Relationships to Other Primes

One-hop neighborhood: parents above, mutual partners to the right, children below.CounterfactualProximity Weightingcomposition: Counterfactual ReasoningCounterfactualReasoning

Parents (1) — more general patterns this builds on

  • Counterfactual Proximity Weighting presupposes Counterfactual Reasoning

    The SIGNAL-GENERATION mechanism that weights a nearby counterfactual into a reward/loss signal by distance; it presupposes counterfactual_reasoning to CONSTRUCT the alternative, then folds it in by proximity. The file: 'reasoning constructs the alternative; proximity weighting weights it into a signal'.

Path to root: Counterfactual Proximity Weighting → Counterfactual Reasoning

Neighborhood in Abstraction Space

Counterfactual Proximity Weighting sits in a sparse region of abstraction space (93rd percentile for distinctiveness): few abstractions share its structure, so a faithful description tends to retrieve it precisely rather than landing on a neighbor.

Family — Causality, Counterfactuals & Logic of Claims (22 primes)

Nearest neighbors

Computed from structural-signature embeddings · 2026-06-14

Not to Be Confused With

The nearest neighbour is counterfactual_reasoning, and the distinction is the prime's own load-bearing scope. Counterfactual reasoning is the cognitive operation of constructing and evaluating what would have happened under different conditions — imagining the road not taken, simulating an alternative, inferring a cause by considering its absence. Counterfactual proximity weighting is not the construction of the alternative but the grading of an internal signal by the distance to it: once a nearby counterfactual of different value exists, the reward, regret, attention, or risk-register update is weighted by a decreasing function of how close that counterfactual was. The decisive added commitment is signal generation as a function of proximity — the model s(o) = α·r(o) + (1 − α)·w(d)·c(o), in which the behavioural update tracks s rather than the realised outcome r, with the classical outcome-only case recovered at α = 1. Counterfactual reasoning supplies the alternative c(o); proximity weighting folds it into the signal by distance d. A practitioner who collapses the two will treat the phenomenon as "people reasoning about alternatives" and miss the prime's sharp, auditable content — that the weight on the counterfactual, and specifically whether it tracks true counterfactual probability or mere perceptual vividness, is the single parameter that separates an informative signal from a maladaptive one (the gambler's chased near-miss, the safety officer's correctly-surfaced near-catastrophe).

Counterfactual proximity weighting is also distinct from anchoring, with which it is confused because both describe a judgement being pulled toward a reference value. Anchoring is the influence of an initial or salient number on a subsequent estimate, where the judgement is adjusted insufficiently away from a starting anchor. Counterfactual proximity weighting grades a signal by how near an alternative outcome was in the relevant state space — symbol distance, causal-chain branching, score margin — not by adjustment from a numerical starting point. The structural difference is the role of distance to a counterfactual versus adjustment from an anchor: anchoring has no counterfactual outcome of different value and no proximity metric, while proximity weighting has no starting estimate being insufficiently adjusted. A negotiation example sharpens the contrast: an offer "just barely" rejected functions as a partial-acceptance signal (proximity weighting — the narrow gap to acceptance grades the update), which is a different mechanism from the same offer's number anchoring the next round's estimate, even though both can operate on the same event. Reading proximity weighting as anchoring loses the counterfactual and the distance metric that are the prime's actual machinery.

A thinner confusion is with attention. Attention is the selective allocation of limited processing to some inputs over others. Counterfactual proximity weighting concerns how an outcome's internal signal is computed — weighting a nearby counterfactual into the reward or loss — which may drive attention downstream (a near-miss recalibrating risk perception and thus what one attends to) but is not itself an allocation of processing. The prime is upstream of attention: it generates the graded signal that downstream attention and behaviour then track. Treating proximity weighting as an attentional phenomenon misplaces it in the processing pipeline and obscures the auditable weight w(d) that is its real lever.

For practitioners the distinctions decide where to intervene. Mistake proximity weighting for counterfactual reasoning and you address how people imagine alternatives, missing the calibration of the weight on them. Mistake it for anchoring and you hunt for a reference number that is not the mechanism. Mistake it for attention and you tune allocation downstream of the signal that is the actual object. Naming counterfactual proximity weighting correctly fixes attention on its single auditable parameter — whether w(d) tracks true counterfactual probability or perceptual vividness — and on the three-direction lever (suppress, surface, or re-ground the counterfactual neighbour) that shifts the response across substrates.

Solution Archetypes

No catalogued solution archetypes reference this prime yet.