Jaime de la Rocha
Wednesday 1st June 2016
Time: 4.00pm
Ground Floor Seminar Room
25 Howland Street, London, W1T 4JG
Across-trial dynamics of stimulus priors in an auditory discrimination task
Just as our experience has its origin in our perceptions, our perceptions are fundamentally shaped by our experience. How does the brain build expectations from experience and how do expectations impact perception?
We aim to understand how neural circuits integrate the recent history of stimuli and rewards in order to generate priors, and how these priors are combined with sensory information to bias decisions. We trained rats in a reaction-time two-alternative forced-choice task with stimuli consisting in a parametric superposition of two amplitude-modulated tones. Rats had to discriminate the dominant tone and seek reward in the associated port. We used partially predictable stimulus sequences that, once their statistics were learned, could be used to generate adaptive priors that maximize the performance. These sequences were introduced by defining Repeating trial blocks, in which the probability to repeat the previous stimulus category was P_rep > 0.5, and Alternating blocks with probability P_rep < 0.5.
We found that animals adapted their behaviour exhibiting a repeating choice bias after correct repetitions and a weaker but reliable alternating bias after correct alternations. The magnitude of this bias built up after each correct response but reset to zero after error trials. Probit regression analyses quantified the impact of recent trial history on choices: both response laterality (Left vs Right) and response transition (Alternation vs. Repetition) tended to impact positively in the upcoming trial choice bias. The weights of this terms went down to almost zero after error trials, confirming the reset observed in psychometric curves. Our findings show that priors show build-up-and-reset dynamics across trials allowing animals to capitalize on the predictability of the stimulus sequence.
Bio
Dr. Jaime de la Rocha received his undergraduate degree in Theoretical Physics from the Universidad Autónoma de Madrid (UAM, 1997) and in Mathematics in the UAM in 1999. He did his PhD in the same university working in the laboratory of Néstor Parga on computational models of synapses and neurons (1999-2003). He then was a post-doctoral fellow in the laboratory of Alex Reyes, at New York University (2004-2008) where he worked performing in vitro electrophysiology. In 2008 he joined the laboratory of Ken Harris, at the Center for Molecular and Behavioral Neuroscience at Rutgers University. In 2010 he earned a Ramon y Cajal position and moved to the Institut d'Investigacions Biomèdiques August Pi i Sunyer (IDIBAPS, Barcelona) where last year he became a permanent Group Leader. His work combines experiments in behaving rodents, electrophysiology, quantitaive analysis and computational network modeling to address questions about perception, decision-making and memory.