Investigating Uncertainty in Fear Conditioning and Extinction Using fMRI

Background: Many autistic people are distressed by inconsistent or changing environments, which may contribute a desire for sameness and to elevated levels of anxiety. There is growing evidence from self- and caregiver-report surveys, that anxiety and repetitive/restricted behaviors and interests in autism are associated with “intolerance of uncertainty” (IU), a cognitive bias towards perceiving situations as ambiguous and uncomfortable. However, there little experimental data that looks directly at the link between IU and anxiety in autism. Improved understanding of the physiological response to fear learning and uncertainty may inform treatment for unique aspects of anxiety in autism.

Objectives: To explore functional brain response to fear learning and extinction under different levels of uncertainty, in autistic and neurotypical adults.

Methods: We compare two separate fMRI studies of classical fear conditioning and extinction that utilized different reinforcement schedules in autistic and neurotypical adults with average to above-average cognitive performance. The unconditioned stimulus in each study was a burst of air to the base of the neck that is startling though not painful. In each experiment, one of two visual shapes was designated as the “safe cue” which predicted no threat (i.e., was never reinforced with the air burst) while the other “threat cue” could be followed by the air burst reinforcement. Experiment 1 (n=39) reinforced the threat cue with the air blast 42% of the time. Experiment 2 (n=57) reinforced the threat cue 100% of the time. In both experiments, the fear acquisition stage was followed by a fear extinction phase in which where the threat and safe stimuli were never reinforced by the air burst. Analyses of fMRI data targeted known regions of interest for fear conditioning and extinction including amygdala, insula, and prefrontal cortex.

Results: In the more uncertain context (42% reinforcement), the autism group showed a significantly decreased differentiation of threat and safe cues in right amygdala functioning. The autism group shows an increased (possibly delayed) amygdala and insula response during the now-safe context of the extinction phase. In the more certain (100% reinforcement experimental context, there were expected main effects for condition (threat > safe) but no group differences or group x condition interactions during fear acquisition. FMRI data likewise showed no group differences or interactions during extinction, though pupillometry—which indexes a much faster physiological response—indicates a possible increase in trial-by-trial arousal in autism during the extinction phase. We also found that brain activation to threat cues was associated with self-report sensory processing measures.

Conclusions: Exposure and extinction are frequent techniques for treating anxiety in neurotypical people. Our results suggest that initial fear learning is likely atypical in autism at least in contexts where the rules or rates of reinforcement may be ambiguous; and report an increased physiological arousal to threat during extinction, which could negatively impact extinction-based treatments. Intolerance of Uncertainty, perhaps driven by atypical sensory processing, is likely an important target for treatment of anxiety in autism and more experimental work is needed to track the mechanisms that link IU to anxiety.