A cognitive pathway to punishment insensitivity

Abstract

Individuals differ in their sensitivity to the adverse consequences of their actions, leading some to persist in maladaptive behaviors. Two pathways have been identified for this insensitivity: a motivational pathway based on excessive reward valuation and a behavioral pathway based on autonomous stimulus-response mechanisms. Here, we identify a third, cognitive pathway based on differences in punishment knowledge and use of that knowledge to suppress behavior. We show that distinct phenotypes of punishment sensitivity emerge from differences in what people learn about their actions. Exposed to identical punishment contingencies, some people (sensitive phenotype) form correct causal beliefs that they use to guide their behavior, successfully obtaining rewards and avoiding punishment, whereas others form incorrect but internally coherent causal beliefs that lead them to earn punishment they do not like. Incorrect causal beliefs were not inherently problematic because we show that many individuals benefit from information about why they are being punished, revaluing their actions and changing their behavior to avoid further punishment (unaware phenotype). However, one condition where incorrect causal beliefs were problematic was when punishment is infrequent. Under this condition, more individuals show punishment insensitivity and detrimental patterns of behavior that resist experience and information-driven updating, even when punishment is severe (compulsive phenotype). For these individuals, rare punishment acted as a "trap," inoculating maladaptive behavioral preferences against cognitive and behavioral updating.

Keywords: compulsivity; individual differences; punishment.

Fig. 1.

Phenotypes of punishment avoidance across Response-CS delays in experiment 1. (A) Conditioned punishment task. (B) Mean (±SEM) punished action preferences for the 3 behavioral phenotypes. Pre = pre-punishment; Rev = post-reveal. (C) Mean (±SEM) punished action (R1) preference across blocks per contiguity group. Contiguity manipulations did not significantly affect avoidance. (D) Composition of clusters by contiguity groups (Top), and vice versa (Bottom). Contiguity group did not significantly determine cluster phenotype. *P < 0.05 single mean t test vs. 0.5 preference.

Fig. 2.

Phenotypes of punishment avoidance across Response-CS probabilities in experiment 2. (A) Conditioned punishment task where the probability of a response yielding a CS was 10%, 20%, or 40%. (B) Mean (±SEM) punished action preference for the 3 behavioral phenotypes. Pre = pre-punishment; Rev = post-reveal. (C) Mean (±SEM) punished action preference across blocks per probability group. Stronger Response-CS probabilities led to greater R1 avoidance. (D) Composition of clusters by probability groups (Top), and vice versa (Bottom). Stronger contingencies drove individuals towards being sensitive, whereas weaker contingencies drove individuals to being Compulsive. *P < .05 single mean t test vs. 0.5 preference.

Fig. 3.

Phenotypes of punishment avoidance under fixed utility in experiment 3. (A) Conditioned punishment task with different Response-CS probability conditions (10% vs. 40%), matched on the utility of the punished response. (B) Mean (±SEM) punished action preference for the 3 behavioral phenotypes. Pre = pre-punishment; Rev = post-reveal. (C) Mean (±SEM) punished action preference across blocks per probability group. Stronger Response-CS probabilities led to greater R1 avoidance. (D) Composition of clusters by probability groups (Top), and vice versa (Bottom). Stronger contingencies drove individuals toward being sensitive, whereas weaker contingencies drove individuals to being compulsive. *P < 0.05 single mean t test vs. 0.5 preference.

Fig. 4.

Underpinnings of punishment avoidance phenotypes pooled from experiments 1 to 3. (A) Click rates per action during non-CS periods across clusters [Sensitive (Sen), Unaware (Unw), Compulsive (Com)]. (B) Average point gain across clusters. (C–E) Value ratings for point outcomes [i.e., reward (Rew), attack (Atk)] (C), CSs (CS+, CS−) (D), and actions (R1, R2) (E). (F–H) Self-reported CS→Attack (F) and Action→CS (G) knowledge, and Action→Attack (H) causal inferences. (I) Action→Attack inferences per action (R1, R2), per block, against predicted attack likelihood based on Action→CS and CS→Attack knowledge (Action→CS→Attack prediction). Inset: individual Action→Attack inferences by Action→CS→Attack predictions with regression lines per cluster (Sensitive = turquoise, Unaware = purple, Compulsive = orange). (J) Relationships between R1:R2 bias in attack inferences, valuations, and behavior (dots: individual datapoints per block across experiments; lines: cluster regression line). (K) Relationship between R1:R2 bias in attack inferences and valuations; R1 was increasingly disliked relative to R2 as attack was attributed to R1 over R2, regardless of cluster. (L) Relationship between R1:R2 bias in attack inferences and behavior; avoidance of R1 over R2 corresponded to attributions of attack to R1 over R2 for Sensitives and Unawares, but not Compulsives. (M) Relationship between R1:R2 bias in valuations and behavior; avoidance of R1 over R2 corresponded to valuation of R2 over R1 for Sensitives and Unawares, but not Compulsives. Data in A–I are means ± SEM.