Mael Lebreton | bioRxiv Science

bioRxiv2016

Optimistic reinforcement learning: computational and neural bases

While forming and updating beliefs about future life outcomes, people tend to consider good news and to disregard bad news. This tendency is supposed to support the optimism bias. Whether this learning bias is specific to \"high-level\" abstract belief update or a particular expression of a more general \"low-level\" reinforcement learning process is unknown. Here we report evidence in favor of the second hypothesis. In a simple instrumental learning task, participants incorporated better-than-expected outcomes at a higher rate compared to worse-than-expected ones. In addition, functional imaging indicated that inter-individual difference in the expression of optimistic update corresponds to enhanced prediction error signaling in the reward circuitry. Our results constitute a new step in the understanding of the genesis of optimism bias at the neurocomputational level.

Neuroscience↗

bioRxiv2016

Revisiting the assessment of inter-individual differences in fMRI activations-behavior relationships.

Characterizing inter-individual differences induced by clinical and social factors constitutes one of the most promising applications of neuroimaging. Paving the way for such applications, neuroimaging studies often report between-group differences in \"activations\" or correlations between such \"activations\" and individual traits. Here we raise cautionary warnings about some of those inter-individual analytic strategies. These warnings become critical when measures of \"activations\" are unstandardized coefficients of regressions between BOLD signal and individual behavior.\n\nFirst, using simple algebraic derivations, we show how inter-individual differences results can spuriously arise from neglecting the statistical relationships which link the ranges of individual BOLD activation and of recorded behavior. We also demonstrate how apparently contradictory results and interpretations may simply arise from the interaction of this scaling issue and the pre-processing of the behavioral variables. Second, using computational simulations, we illustrate how this issue percolates the booming field of model-based fMRI. Finally, we outline a set of recommendations, which might prove useful for researcher and reviewers confronted with questions involving inter-individual differences in neuroimaging.\n\nAuthor SummaryCharacterizing inter-individual differences induced by clinical and societal factors constitutes one of the most promising applications of neuroimaging. Paving the way for such applications, an increasing fraction of neuroimaging studies reports between-group differences in \"activations\" or correlations between \"activations\" and individual traits. In this manuscript, we focus on the typical analytical strategies employed in studies investigating how differences in behavior between individuals or groups of individuals are translated to differences of activations in specific brain regions. We notably question whether they are suitable to support inferences and claims about the neural underpinnings of differential cognition. Our core results demonstrate that typical inter-individual results can spuriously arise by overlooking plausible statistical relationships that link the ranges of individual BOLD activation with the ranges of produced behavior. We argue that these results challenge current classical interpretations of inter-individual results. Highlighting the methodological and theoretical gaps regarding the analysis and interpretation of inter-individual differences is fundamental to fulfilling the promises of neuroimaging.

Neuroscience↗