Title: Néstor Parga, Autonomous University of Madrid
Start date: Nov 03 2017
Start time: 11:15 am
End time: 12:15 pm
Location: Max Planck House Lecture Hall

Abstract: Conditioning experiments and theoretical studies support the hypothesis that the firing activity of dopamine (DA) neurons codes for the error in the prediction of reward and acts as a teaching signal used to guide choices [1]. However in real life situations sensory cues are often ambiguous and the role of DA in perceptual decision-making tasks is poorly understood. To obtain some insight into this issue we investigated the DA firing activity in two tasks in which monkeys had to make decisions based on vibrotactile stimulation.  In one of them the animals were engaged in detecting weak stimuli delivered at uncertain times [2]. In the other, they had to discriminate between the frequencies of two stimuli that were changed from trial to trial [3]. Using a combination of data analysis and computational modeling we analyzed the role of DA neurons in processes that include detection of weak stimuli, time estimation, stimulus comparison, contraction bias and working memory. 

In both tasks the response of the neurons to a reward-predicting cue coded the certainty about the future choice. In the discrimination task, we found that the DA response to the stimuli could be explained in terms of inference processes that lead to the formation of the contraction bias, a well-known phenomenon present in 2AFC tasks by which the frequency of the first stimulus is treated as if its value were shifted to the center of its range. Finally, after the application of the first stimulus the population activity of DA neurons did not depend on the value of its frequency and increased throughout the delay period. 

Our results support the notion that under uncertain sensory conditions the DA activity reflects neural computations related to the elaboration of the decision and that, at the same time, the phasic activity of DA neurons codes reward prediction errors.

[1] Schultz W, Dayan P, Montague PR (1997). A neural substrate of prediction and reward. Science 275(5306):1593–1599.
[2] Sarno S, de Lafuente V, Romo R and Parga N (2017). The dopamine reward prediction error signal codes the temporal evaluation of a perceptual decision report. To appear in PNAS.
[3] Sarno S, Romo R, Parga N et al. (2017). In preparation.

Host: Daniel Zaldivar

More Information