Biology subjects

Vidal, Y.

Publications and source records attributed to Vidal, Y..

5 recordsLinked to original sources

bioRxiv2026

Predictive perception via simultaneous learning and inference

Perception has been proposed to involve an inferential process that combines noisy sensory evidence with prior expectations to estimate the latent states of the environment. Exact Bayesian inference is intractable in the continuous, high-dimensional state spaces of the natural world. Variational schemes address this by restricting the form of the posterior, whereas sampling schemes represent the posterior with a finite set of samples. In both cases the approximation concerns the posterior rather than the state space over which it is defined. Here we place the approximation elsewhere. Rather than restricting the form of the posterior, we restrict the state space, modeling environmental states as discrete and thereby making Bayesian filtering exact. Beliefs are then unconstrained in shape and what has to be learned can be reduced to a transition matrix updated online by a local rule. The cost is that the states, rather than the distribution, must be specified in advance. The Simultaneous Learning and Inference Model (SLIM) combines this exact filtering with a gated Hebbian rule that learns transitions from inferred rather than observed states. In simulations SLIM recovers environmental dynamics under sensory noise and adapts when those dynamics change, with error growing only once the sensor becomes uninformative. A hierarchical instantiation reproduces local and global prediction error effects. Applied to two auditory decision-making experiments under noise, SLIM reproduces behavioral signatures of expectation and supplies trial-level measures of expectation and surprise. Exact inference over a small discrete state space with a single local Hebbian rule is therefore sufficient to account for these phenomena, without an explicit optimization objective and without a parametric approximation to the posterior.

neuroscience↗

bioRxiv2023

An adversarial collaboration to critically evaluate theories of consciousness

Different theories explain how subjective experience arises from brain activity1,2. These theories have independently accrued evidence, yet, confirmation bias and dependence on design choices hamper progress in the field3. Here, we present an open science adversarial collaboration which directly juxtaposes Integrated Information Theory (IIT)4,5 and Global Neuronal Workspace Theory (GNWT)6-10, employing a theory-neutral consortium approach11,12. We investigate neural correlates of the content and duration of visual experience. The theory proponents and the consortium developed and preregistered the experimental design, divergent predictions, expected outcomes, and their interpretation12. 256 human subjects viewed suprathreshold stimuli for variable durations while neural activity was measured with functional magnetic resonance imaging, magnetoencephalography, and electrocorticography. We find information about conscious content in visual, ventro-temporal and inferior frontal cortex, with sustained responses in occipital and lateral temporal cortex reflecting stimulus duration, and content-specific synchronization between frontal and early visual areas. These results confirm some predictions of IIT and GNWT, while substantially challenging both theories: for IIT, a lack of sustained synchronization within posterior cortex contradicts the claim that network connectivity specifies consciousness. GNWT is challenged by the general lack of ignition at stimulus offset and limited representation of certain conscious dimensions in prefrontal cortex. Beyond challenging the theories themselves, we present an alternative approach to advance cognitive neuroscience through a principled, theory-driven, collaborative effort. We highlight the challenges to change peoples mind 13 and the need for a quantitative framework integrating evidence for systematic theory testing and building.

neuroscience↗

bioRxiv2022

Oculomotor and neural correlates of semantic and morphological priming in natural sentence reading

AO_SCPLOWBSTRACTC_SCPLOWOur current understanding of visual word identification is difficult to extend to text reading-both experiments and theories focus primarily, if not exclusively, on out-of-context individual words. Here, we try to fill this gap by studying cross-word semantic and morphological priming within sentences in natural reading, in a novel coregistration paradigm with simultaneous recording of eye movements and electroencephalography. We report results from both eye tracking measures, and, more importantly, from Fixation-Related potentials, time-locked to the fixation onset on the target word. In both, semantic facilitation clearly emerged, while we observed no effect of morphological priming. These results may indicate that morphological agreement is at least partially computed outside of the lexical-semantic system which gives rise to semantic priming. These results provide new insight into the neural correlates of semantic and morphological priming in natural reading, revealing lexical dynamics as they likely emerge in our everyday reading experience.

neuroscience↗

bioRxiv2020

A general-purpose mechanism of visual feature association at the core of word learning

Writing systems are a recent cultural invention, which makes it unlikely that specific cognitive mechanisms have developed through selective pressure for reading itself. Instead, reading might capitalize on evolutionary older mechanisms that originally supported other tasks. Accordingly, animals such as baboons can be trained to perform visual word recognition. This suggests that the visual mechanisms supporting reading might be phylogenetically old and domain-general. Here we propose that if the human reading system relies on domain-general visual mechanisms, effects that are typically found within the domain of reading should also be observable with non-orthographic visual stimuli. To test this hypothesis, we systematically tested different types of visual material with the same experimental design. Subjects were passively familiarized with a set of composite visual items, and then tested in an oddball paradigm for their ability to detect novel stimuli. Some of these novel stimuli shared their statistical structure with the familiar items, and were found to be hard to detect in two experiments using strings of letter-like symbols; this replicates the well-known, and supposedly reading-specific, bigram effect. Crucially, in two further experiments we show that the same effect emerges with made-up, 3D objects and sinusoidal gratings. The effect size was equivalent across experiments, despite the use of radically different stimuli. These data suggest that a fundamental mechanism behind visual word learning also supports the learning of other visual stimuli, implying that such mechanism is general-purpose. This mechanism would enable the statistical learning of regularities in the visual environment.

neuroscience↗

bioRxiv2019

Predictive Coding beyond simple tone-based oddball designs

As the evidence of predictive processes playing a role in a wide variety of cognitive domains increases, the brain as a predictive machine becomes a central idea in neuroscience. In auditory processing a considerable amount of progress has been made using variations of the Oddball design, but most of the existing work seems restricted to predictions based on physical features or conditional rules linking successive stimuli. To characterise the predictive capacity of the brain to abstract rules, we present here two experiments that use speech-like stimuli to overcome limitations and avoid common confounds. Pseudowords were presented in isolation, intermixed with infrequent deviants that contained unexpected phoneme sequences. As hypothesized, the occurrence of unexpected sequences of phonemes reliably elicited an early prediction error signal. These prediction error signals do not seemed to be modulated by attentional manipulations due to different task instructions, suggesting that the predictions are deployed even when the task at hand does not volitionally involve error detection. In contrast, the amount of syllables congruent with a standard pseudoword presented before the point of deviance exerted a strong modulation. Prediction errors amplitude doubled when two congruent syllables were presented instead of one, despite keeping local transitional probabilities constant. This suggest that auditory predictions can be built integrating information beyond the immediate past. In sum, the results presented here further contribute to the understanding of the predictive capabilities of the human auditory system when facing complex stimuli and abstract rules.\n\nSignificance StatementThe generation of predictions seem to be a prevalent brain computation. In the case of auditory processing this information is intrinsically temporal. The study of auditory predictions has been largely circumscribed to unexpected physical stimuli features or rules connecting consecutive stimuli. In contrast, our everyday experience suggest that the human auditory system is capable of more sophisticated predictions. This becomes evident in the case of speech processing, where abstract rules with long range dependencies are universal. In this article, we present two electroencephalography experiments that use speech-like stimuli to explore the predictive capabilities of the human auditory system. The results presented here increase the understanding of the ability of our auditory system to implement predictions using information beyond the immediate past.

neuroscience↗