Search

Normative theory predicts that feedback should not affect decisions under risk, but past findings disagree. Here, the authors show that feedback shifts risk-taking by changing attitudes rather than through learning.

Reinforcement learning task-based behavioral and computational measures displayed low test–retest reliability at the individual level. Also in contrast to self-assessed personality measures, behavioral and computational measures were poor predictors of mental health measures, representing a challenge for computational psychiatry.

Lefebvre et al. present behavioural and neural evidence showing that the ‘optimism bias’ is a manifestation of a general cognitive tendency for preferential learning from positive, compared with negative, outcomes.

In contrast to predictions from learning theory, humans learn to seek rewards and avoid punishments equally well. Here the authors offer an elegant solution to this problem by demonstrating that humans learn option values relative to a reference point subserved by a common neural substrate.

The functional role of the human thalamus in reinforcement learning is debated. Here, using intra-thalamic recordings in humans, the authors report that thalamic low-frequency oscillations correlate with variables for learning from both reward and punishment.

Anlló et al. show that context effects on reward value encoding are found across 11 diverse countries, suggesting that this may be a basic feature of human decision-making.

Standard decision models assume that all options' values are encoded on a common scale by a unique representation system. Across nine experiments, Garcia et al. provide evidence that challenges this assumption: participants treat experiential and symbolic options asymmetrically.

We know from past research that humans tend to learn more from good news than from bad news. Chambon et al. show that this confirmation bias is specific to free choices (that is, feedback towards actions that reflect people’s own decision).

Humans often make sub-optimal decisions, choosing options that are less advantageous than available alternatives. Using computational modeling of behavior, the authors demonstrate that such irrational choices can arise from context dependence in reinforcement learning.

The mechanism of confidence formation in learning remains poorly understood. Here, the authors show that both dorsal and ventral prefrontal networks encode confidence, but only the ventral network incorporates the valence-induced bias.

Findling, Skvortsova et al. find that a large fraction of non-greedy decisions that humans make in volatile environments do not stem from exploration but from the limited precision of learning, and further identify its neurophysiological correlates.

Though it's important to influence others' decisions, the neural correlates of persuasive strategies are not known. Here, authors show that people change their advice based on its accuracy and whether they are being listened to, and identify the distinct brain regions underpinning each strategy.

In the 1930s, philosopher John Dewey stated: “We do not learn from experience… we learn from reflecting on experience.” The question of how we learn from the consequences of our actions has been investigated for decades. When deliberating between options, it is assumed that the outcome of our choice is used as a feedback signal to learn the value of the chosen option. But what about the forgone alternative? In a recent paper, Biderman and Shohamy show that we also revise the valuation of forgone options, assuming them to be inversely related to that of chosen ones.

Some large language models show reasoning errors akin to humans in cognitive bias tasks. However, humans and models respond differently to prompting strategies, highlighting differences in cognitive processing.

This study shows that, following social isolation, adolescents show increased sensitivity to rewards during effort-based decision making and reward learning. Access to virtual social interactions during isolation partially remediates these effects.

This perspective discusses the theoretical and statistical foundations of the analyses of inter-individual differences in task-related functional MRI, offering recommendations for improving statistical validity and interpretability of inter-individual differences in functional MRI.

Behavioral results suggest that learning by trial-and-error (i.e., reinforcement learning) relies on a teaching signal, the prediction error, which quantifies the difference between the obtained and the expected reward. Evidence suggests that distinct cortico-striatal circuits are recruited to encode better-than-expected (positive prediction error) and worst-than-expected (negative prediction error) outcomes. A recent study by Villano et al.¹ provides evidence for differential networks that underlie learning from positive and negative prediction errors in humans using real-life behavioral data. More specifically, they found that university students are more likely to update beliefs concerning grade expectations following positive rather than negative prediction errors.

In non-competitive settings, learners manifest a preference for broadcasting useful experience-based information to other learners, even when it comes at a social or economic cost.

Multivariate neuroimaging analyses of participants in a texture discrimination task suggests that cognitive fatigue via repeated stimulus exposure impacts the activity of task-relevant networks and is associated with tangible behavioral repercussions.