Neural and hormonal systems underlying human reward-seeking behavior
Our evolutionary history has endowed us with biological systems for identifying those elements of the environment that contribute to our biological fitness and for modifying our behaviors to allow us to acquire them.
Theoretical propositions suggest that the ability to detect changes in the statistics underlying in the environment may be useful for rapidly adapting our behaviors. However little is known about the neural representation of the quantity representing the evidence for a change point, unexpected uncertainty. In Chapter 2, I describe a study in which humans interact with an unstable reward environment while undergoing fMRI. Representations of unexpected uncertainty were found in multiple cortical areas, as well as the noradrenergic brainstem nucleus locus coeruleus. Other unique cortical regions were found to encode estimation uncertainty, or the uncertainty in ones estimates of the reward contingencies, and risk, or ones estimate of the stochasticity of the environment. Collectively, these findings support theoretical models in which uncertainty computations determine the speed of learning.
Although learning from direct experience in this way is vital to our survival, humans are also particularly adept at learning from conspecifics. However, it is not known whether differing computational strategies thought to support experiential learning, model-based and model-free learning, also support learning by observation. Chapter 3 describes a study in which human participants played a multi-armed bandit task that encouraged them to employ both experiential and observational learning while they underwent fMRI. Model-based learning signals are found during both observational and experiential learning in the intraparietal sulcus. However, unlike in experiential learning, model-free learning signals in the ventral striatum were not detectable during observational learning. These results provide insight into the flexibilty of the model-based learning system, and further suggest that the model-free learning system may be less flexible with regard to its involvement in observational learning.
While Chapters 2 and 3 are concerned with modifying reward-seeking behavior in reponse to changes in the external environment, Chapter 4 examines a modification of reward-seeking behavior in response to changes in the internal hormonal environment. Specifically, it describes how the behavior of human males in a simple economic game was influenced by the administration of testosterone. Although a popular view on the role of testosterone in human social behaviour proposes that it increases aggression, a recent theory states that it instead promotes behaviors that enhance social status. In a double-blind, placebo-controlled between-subjects design, administration of testosterone increased punishment of players who treat the participants unfairly but also increased reward of those who treat them generously. Our findings are inconsistent with the view that testosterone simply increases aggression and provides causal evidence for the social-status hypothesis in men.
In Chapter 5, I describe an investigation of the phenomenon of 'choking under pressure', in which reward-seeking behavior is compromised by the promise of high reward for successful performance. A novel approach to attenuating such 'choking under pressure' using cognitive reappraisal of the incentive is described and tested. When participants performed a demanding motor task under reappraisal, choking was indeed significantly reduced, with the magnitude of this reduction being predicted by the striatal BOLD response to incentive magnitude. In addition, application of the reappraisal strategy was associated with reduced sympathetic arousal during trials on which performance failed at high levels of incentive. These results suggest that reappraisal of the incentive is indeed a promising intervention for attenuating choking under pressure.