Threat of shock and aversive inhibition: Induced anxiety modulates Pavlovian-instrumental interactions

Abstract

Anxiety can be an adaptive response to potentially threatening situations. However, if experienced in inappropriate contexts, it can also lead to pathological and maladaptive anxiety disorders. Experimentally, anxiety can be induced in healthy individuals using the threat of shock (ToS) paradigm. Accumulating work with this paradigm suggests that anxiety promotes harm-avoidant mechanisms through enhanced inhibitory control. However, the specific cognitive mechanisms underlying anxiety-linked inhibitory control are unclear. Critically, behavioral inhibition can arise from at least 2 interacting valuation systems: instrumental (a goal-directed system) and Pavlovian (a "hardwired" reflexive system). The present study (N = 62) replicated a study showing improved response inhibition under ToS in healthy participants, and additionally examined the impact of ToS on aversive and appetitive Pavlovian-instrumental interactions in a reinforced go/no-go task. When Pavlovian and instrumental systems were in conflict, ToS increased inhibition to aversive events, while leaving appetitive interactions unperturbed. We argue that anxiety promotes avoidant behavior in potentially harmful situations by potentiating aversive Pavlovian reactions (i.e., promoting avoidance in the face of threats). Critically, such a mechanism would drive adaptive harm-avoidant behavior in threatening situations where Pavlovian and instrumental processes are aligned, but at the same time, result in maladaptive behaviors when misaligned and where instrumental control would be advantageous. This has important implications for our understanding of the mechanisms that underlie pathological anxiety. (PsycINFO Database Record

Figure 1

Trial sequence for all the action–valence (A-V) conditions in the reinforced go/no-go task. Each A-V condition began with a neutral block (36 trials) to allow reaction times (RTs) to equilibrate, followed by one of the four A-V experimental conditions (36 trials). Displayed are the trial sequence for each A-V condition under (a) safe and (b) threat. The safe and threat blocks were presented in alternating order, counterbalanced across participants. See the online article for the color version of this figure.

Figure 2

The action–outcome reinforcement schedules for each action–valence task condition. (a) In the go win (GW) condition, actions were biased toward go, by rewarding correct go responses more strongly than correct no-go responses; incorrect responses were not rewarded. (b) In the go avoid (GA) condition actions were biased toward go, by punishing incorrect no-go responses more harshly than incorrect go responses; correct responses were not punished. (c) In the no-go win (NGW) condition actions were biased toward no-go, by rewarding correct no-go responses more strongly than correct go responses; incorrect responses were not rewarded. (d) In the no-go avoid (NGA) condition actions were biased toward no-go, by punishing incorrect go responses more harshly than incorrect no-go responses; correct responses were not punished.

Figure 3

Violin and overlaid box and error bar plots of the sustained attention to response task data. (a) Accuracy on the no-go trials across threat and safe conditions. Threat significantly increased accuracy compared with safe blocks (* p = .001). (b) Reaction times on go trials across threat and safe conditions. Threat significantly slowed responses (* p = .019). Black dots represent the mean and associated error bars represent standard error of the mean for within-subjects variance.

Figure 4

Violin plots with overlaid box and error bar plots showing normalized RTs (standardized against the practice baseline) for correct go trials in each action–valence condition under threat and safe conditions. Less negative values indicate slower responses. Threat significantly slowed responses during go to avoid punishment compared with safe blocks (* p = .04). This indicates that threat selectively potentiates inhibition of actions in the face of punishment when the Pavlovian and instrumental systems are in conflict, by increasing reliance on aversive Pavlovian biases. Black dots depict the mean and associated error bars represent standard error of the mean for the within-subjects variance.