AMYGDALA CONTRIBUTES TO LOSS PERCEPTION IN A RODENT REFLECTION EFFECT TASK

Mathematical models can outperform human specialists by reducing cognitive bias, though performance gains are usually modest. Despite extensive artificial intelligence research, the neural basis of human cognitive biases has received relatively little attention. One such bias is the reflection effect, a well-established decision-making phenomenon in which individuals prefer safe options in gain contexts but become risk-seeking when facing potential losses. This effect appears evolutionary conserved and has recently been demonstrated in non-human primates; however, its neural and neurochemical basis remains largely unexplored.
Here, we investigated whether rats exhibit the reflection effect and aimed to identify the underlying neural circuits and neuromodulatory mechanisms. First, we showed that reward certainty and delay similarly influence rodent decision-making, suggesting that delayed rewards acquire loss-like value. Based on this, we developed a novel rodent reflection task dissociating gain and loss domains.
Using photometry, we measured dopamine release in the nucleus accumbens and neural activity in the basolateral amygdala during task performance. We found that wins were consistently encoded as high-value outcomes and losses as low-value outcomes in both gain and loss contexts. Despite this stable value encoding, rats dynamically shifted from risk-averse behavior in the gain domain to risk-seeking behavior in the loss domain within the same session.
Finally, optogenetic silencing of the basolateral amygdala during loss collection increased risk-seeking behavior following losses. These results indicate that the amygdala is important for loss perception and shifts in risk preference, establishing rodents as a suitable model for investigating the neural mechanisms of the reflection effect.