From Learning to Choosing: How Decision-Making Evolves with Experience in Rats

Decision-making is a fundamental process that guides actions by selecting between various options based on their known or presumed outcomes, often using sensory inputs (Carandini and Churchland, 2013).The neural mechanisms by which the brain integrates complex information to make decisions are typically studied by measuring neural recordings, response times, and choice accuracy using well-trained animals (Carandini and Churchland, 2013).These experiments often employ two-alternative forced-choice (2AFC) designs, where animals are trained to choose between two stimuli presented simultaneously, assuming that the animal learns the value of each stimulus and decides based on an internal comparative evaluation.However, the traditional 2AFC design may be limited, as it does not consider that stimuli are not encountered simultaneously in nature and that decision-making strategies evolve during learning (White et al., 2024).Kacelnik et al. ( 2011) highlight these concerns, arguing that animals' choices on a 2AFC task could be predicted by latencies observed when animals are provided with a single offer.Furthermore, they suggest that the act of deliberation, observed when animals slow their response times to make a choice, could be an artifact produced as animals learn the 2AFC.The recent study by White et al. (2024) in eNeuro aimed to address these limitations by investigating decision-making dynamics during the initial learning phase.By isolating learning values of individual stimuli from the decision-making process, the researchers sought to understand how male rats can make choices for the first time on a 2AFC and how those choices change with experience.The behavioral experiment comprised two main stages: value learning and choice learning (Fig. 1).In the value learning stage, rats were introduced to a single stimulus of either high or low luminance, where nose pokes at the port below the stimulus led to the delivery of a high (16% sucrose) or low (4% sucrose) reward, respectively.The choice learning stage consisted of single-offer trials for two-thirds of the total trials, where rats were presented with either the high or low-luminance cue.The remaining one-third of trials were dual-offer trials, where animals were simultaneously offered both high-and low-luminance stimuli, randomized by side.Rats' response latencies and choice percentages were measured.During the initial stages of choice learning, rats consistently preferred the highluminance stimuli.Median response latencies for dual-offer trials were greater compared with single offers and were greater for single-offer trials with low-value stimuli compared with those with high-value stimuli.These latency differences were most pronounced in the first session but persisted over the remaining sessions.These differences indicate that even after dissociating the value learning stage from the decision-making process, rats showed evidence of deliberation, which remained present with experience.To examine changes in the response time distribution that occur with experience, the authors utilized an ExGauss fitting.In this model, response times are fitted with Gaussian and exponential components, considering the peak and tail of the response time distribution, respectively.This enables the quantification of sensorimotor processing (Gaussian component) and variability (exponential component) within the data.The fitting showed that when choosing the high-value reward, there was increased variability in

Récupéré en direct depuis OpenAlex et désinversé. Les résumés ne sont pas conservés dans cette base de données : les index inversés représentent 8,6 Go des 9,3 Go de texte de la base, et le serveur dispose de 13 Go libres.