APAlexis Pérez-Bellido

Alexis Pérez-Bellido

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Alexis Pérez-Bellido has a total of 37 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Metacognitive Efficiency Reduces Confirmation Bias in Perceptual Decision Making

    by Alexis Perez-Bellido, Ruben Moreno-Bote, Lluis Fuentemilla

    Humans exhibit a pervasive drive toward self-consistency, often failing to revise previous decisions even when confronted with contradictory evidence. Here, we investigate the computational mechanisms underlying decision revision in perceptual tasks, examining the regulatory role of metacognition. To do so, we capitalize on a novel paradigm in which participants are repeatedly presented with identical sensory information and allowed to revise their choices after each exposure. Our results reveal that repeated exposure to the same stimulus systematically biases subsequent judgments toward prior responses. Using drift-diffusion modeling, we tested competing explanations incorporating different assumptions about how prior choices affect evidence accumulation. Our findings indicate that consistency biases emerge from asymmetric sensory weighting, selectively amplifying information consistent with previous choice (a phenomenon akin to confirmation bias). Crucially, individuals with higher metacognitive skills exhibited weaker confirmatory biases and more flexible integration of repeated sensory information, enabling greater adaptability in decision-making. These findings highlight the continuous nature of perceptual inference and underscore metacognition's pivotal role in mitigating bias and optimizing decision flexibility.

  • Open Access

    Predictable modality transitions and meaningful stimuli facilitate sequential statistical learning between sensory modalities.

    by Alexis Pérez-Bellido, Daniel Duato, Francesco Giannelli, Ruth de Diego-Balaguer

    Statistical learning is the process of extracting regularities from the environment. Although sequential cross-modal regularities are ubiquitous (e.g., the sight of a falling ball followed by the sound of impact), previous research has failed to demonstrate that such learning occurs between sensory modalities. This has led to the view that sequential statistical learning is modality specific. The present study investigates factors that may determine statistical learning of cross-modal sequences. In the first experiment, participants viewed a stream of meaningless visual fractals and synthetic auditory stimuli. The sequence of stimuli could be grouped into either unimodal or cross-modal pairs based on their transitional probabilities. Using implicit and explicit measures of learning, we found that participants learned only the unimodal pairs. In the second experiment, pairs were presented in separate unimodal and cross-modal blocks. The cross-modal blocks alternated between visual and auditory modalities, allowing participants to anticipate the upcoming modality. This manipulation enabled significant implicit statistical learning for the cross-modal pairs. This suggests that the predictability of modality transitions facilitates appropriate deployment of attention across sensory modalities, which is crucial for learning cross-modal sequential contingencies. In the third experiment, we used audiovisual stimuli with semantic content. Here, participants were able to implicitly learn and explicitly recognize statistical regularities between cross-modal pairs even when the upcoming modality was unpredictable. Together, these findings challenge the view that sequential statistical learning is strictly modality specific, showing that it occurs when sensory-level limitations are bypassed by attentional cues or when learning engages higher level semantic representations shared across modalities. (PsycInfo Database Record (c) 2026 APA, all rights reserved).


  • Open Access

    Attention and Explicit Knowledge Drive Predictive Sharpening in Early Visual Cortex

    by Marc Sabio-Albert, David Richter, Lluís Fuentemilla, Alexis Pérez-Bellido

    Perception is increasingly understood as an inferential process, whereby what we perceive results from the integration of sensory inputs with expectations derived from prior knowledge. Top-down predictions have been shown to alter the encoding of sensory information, from early to late stages of processing. Yet, how such predictions shape neural representations in sensory cortices remains debated. Competing accounts suggest that predictions either sharpen neural representations by enhancing selectivity or dampen activity by broadly suppressing stimulus-driven responses. In a preregistered fMRI study, we tested whether these effects depend on the level of attentional engagement and explicit knowledge of predictive associations. Using a multisensory fMRI paradigm with concurrent but independent visual and auditory probabilistic associations (75% validity) and manipulated attention, we investigated predictive effects in human early visual cortex. Consistent with prior work, expected visual stimuli elicited reduced BOLD activity. Critically, sharpening of expected visual stimuli occurred exclusively when visual inputs were attended and the concurrently presented auditory inputs expected. In addition, the magnitude of the sharpening of visual representations correlated positively with participants’ explicit knowledge of the visuo-predictive associations. These findings highlight the key roles of attention and explicit knowledge in promoting predictive sharpening and underscore the need to study predictive processing in more ecologically valid, multisensory contexts.

  • Open Access

    Anticipating Multisensory Environments: Revealing Interactions in Predictions Across Sensory Modalities Unveils the Supra-modal Nature of the Predictive System

    by Marc Sabio-Albert, Lluís Fuentemilla, Alexis Pérez-Bellido

    Our experience is generally framed in multisensory environments abundant in predictive information. Accordingly, previous research on statistical learning has shown that humans can learn regularities in different sensory modalities in parallel, but left open the question of whether participants also anticipated multisensory inputs. Across two experiments, we tested whether participants capitalized on prior expectations to facilitate perceptual processing of concurrent but independent multisensory stimuli. Participants were exposed to concurrent pairs of auditory and visual low-level stimuli (i.e., gratings and tones). In each experimental block, participants had to attend to one modality and discriminate whether the second stimulus of the attended pairs (targets) were deviant or standard, while ignoring stimuli from the other sensory modality (distractors). Orthogonal to the task goal, both the attended and unattended pairs followed transitional probabilities, so targets and distractors could be expected or unexpected. Participants performed better for expected compared to unexpected targets. This effect generalized to the distractors but only when targets were expected. Such interactive effects suggest that predictions may be gated by a supra-modal system with shared resources across sensory modalities that are distributed according to their respective behavioral relevance.

  • Open Access

    Measuring phase-amplitude coupling opposition in neurophysiological signals with the Mean Opposition Vector Index (MOVI)

    by Ludovico Saint-Amour di Chanaz, Alexis Pérez-Bellido, Lluís Fuentemilla

    Background The phase–amplitude coupling (PAC) opposition between distinct neural oscillations is critical to understanding brain functions. Available methods to assess phase-preference differences between conditions rely on density of occurrences. Other methods like the Kullback-Leibler Divergence (DKL) assess the distance between two conditions by transforming neurophysiological data into probabilistic distributions of phase-preference and assessing the distance between them. However, these methods have limitations such as susceptibility to noise and bias. New Method We propose the “Mean Opposition Vector Index” (MOVI), a parameter-free, data-driven algorithm for unbiased estimation of PAC opposition. MOVI establishes a unified framework that integrates the strength of PAC to account for reliable unimodal differences in phase-specific amplitude coupling between neurophysiological datasets. Results We found that MOVI accurately detected phase opposition, was resistant to noise and gave consistent results with low or asymmetrical number of trials, therefore in conditions more similar to experimental studies. Comparison with existing methods MOVI outperformed Jensen-Shannon Divergence (JSD), an adaptation of the DKL, in terms of sensitivity, specificity, and accuracy to detect phase opposition. Conclusions MOVI provides a novel and useful approach to study of phase-preference opposition in neurophysiological datasets.

  • Open Access

    Predictable Modality Transitions and Amodal Representations Enable Crossmodal Statistical Learning

    by Alexis Pérez-Bellido, Daniel Duato, Francesco Giannelli, Ruth de Diego-Balaguer

    Statistical learning involves extracting statistical regularities from the environment. Despite the ubiquitous crossmodal relations in our environment, previous research has failed in showing crossmodal learning suggesting that statistical learning is modality-specific, occurring within but not between sensory modalities. The present study investigates under which circumstances statistical learning can occur between modalities. In the first experiment, participants viewed a stream of meaningless visual fractals and synthetic auditory stimuli. Importantly, the sequence of stimuli could be grouped into unimodal or crossmodal pairs based on their internal transitional probabilities. Using implicit and explicit measures of learning, we found that participants only learned the unimodal pairs. In the second experiment, pairs were presented in separate unimodal and crossmodal blocks. The crossmodal blocks alternated visual and auditory modalities allowing participants to anticipate the upcoming modality. This manipulation allowed significant statistical learning for the crossmodal pairs, reflected only by implicit measures. This suggests that modality transitions predictability aids correct attention deployment across sensory modalities, crucial for learning crossmodal statistical contingencies. In the third experiment, where audiovisual stimuli with semantic content were used leaving access to an amodal shared representation, participants could explicitly recognize statistical regularities between crossmodal pairs even when the upcoming modality was unpredictable. This finding suggests statistical learning between crossmodal pairs can occur when sensory-level limitations are bypassed, and when learning unfolds at an amodal level of representation. These findings challenge the view that statistical learning is strictly modality-specific, instead indicating that crossmodal statistical learning depends on attentional mechanisms and representational level. Public significance statement: Previous research has established that low-level statistical learning occurs within individual sensory modalities. However, evidence for statistical learning across modalities remains limited. This has led to the hypothesis that statistical learning operates as a modality-specific system, constrained by the perceptual properties and content of each sensory system. In this study, we demonstrate that modality transitions’ predictability enables crossmodal statistical learning, even in the absence of semantic information. Notably, this constraint does not apply to meaningful stimuli, which can be processed at an amodal level. Our findings broaden the classical neurobiological model of statistical learning by highlighting the roles of attentional mechanisms and representational levels in supporting crossmodal statistical learning.


  • Open Access

    Magnetoencephalography recordings reveal the neural mechanisms of auditory contributions to improved visual detection

    by Alexis Pérez-Bellido, Eelke Spaak, Floris P. de Lange

    Sounds enhance the detection of visual stimuli while concurrently biasing an observer’s decisions. To investigate the neural mechanisms that underlie such multisensory interactions, we decoded time-resolved Signal Detection Theory sensitivity and criterion parameters from magneto-encephalographic recordings of participants that performed a visual detection task. We found that sounds improved visual detection sensitivity by enhancing the accumulation and maintenance of perceptual evidence over time. Meanwhile, criterion decoding analyses revealed that sounds induced brain activity patterns that resembled the patterns evoked by an actual visual stimulus. These two complementary mechanisms of audiovisual interplay differed in terms of their automaticity: Whereas the sound-induced enhancement in visual sensitivity depended on participants being actively engaged in a detection task, we found that sounds activated the visual cortex irrespective of task demands, potentially inducing visual illusory percepts. These results challenge the classical assumption that sound-induced increases in false alarms exclusively correspond to decision-level biases.


  • Open Access

    Seeing sounds: Neural mechanisms underlying auditory contributions to visual detection

    by Alexis Pérez-Bellido, Eelke Spaak, Floris P. de Lange

    Sounds enhance the detection of visual stimuli while concurrently biasing an observer’s decisions. To investigate the neural mechanisms that underlie such multisensory interactions, we decoded time-resolved signal detection theory (SDT) sensitivity and criterion parameters from neural activity using magnetoencalography, while participants performed a visual detection task. Using temporal generalization analysis, we found that sounds improve visual detection by enhancing the maintenance of the most informative perceptual samples over time. In parallel, criterion decoding analyses revealed that sounds evoke patterns of activity that resembled the patterns evoked by an actual visual stimulus. These two complementary mechanisms of audiovisual interaction differed in terms of their automaticity: Whereas the sound-induced enhancement in visual information maintenance depended on participants being actively engaged in a detection task, sounds evoked visual activity patterns in the visual cortex in a bottom-up fashion, challenging the classical assumption that sound- induced reductions in criterion correspond to decision-level biases.


  • Open Access

    Amodal completion instead of predictive coding can explain activity suppression of early visual cortex during illusory shape perception

    by Chuyao Yan, Alexis Pérez-Bellido, Floris P. de Lange

    A set of recent neuroimaging studies observed that the perception of an illusory shape can elicit both positive and negative feedback modulations in different parts of the early visual cortex. When three Pac-Men shapes were aligned in such a way that they created an illusory triangle (i.e., the Kanizsa illusion), neural activity in early visual cortex was enhanced in those neurons that had receptive fields that overlapped with the illusory shape but suppressed in neurons whose receptive field overlapped with the Pac-Men inducers. These results were interpreted as congruent with the predictive coding framework, in which neurons in early visual cortex enhance or suppress their activity depending on whether the top-down predictions match the bottom-up sensory inputs. However, there are several plausible alternative explanations for the activity modulations. Here we tested a recent proposal (Moors, 2015) that the activity suppression in early visual cortex during illusory shape perception reflects neural adaptation to perceptually stable input. Namely, the inducers appear perceptually stable during the illusory shape condition (discs on which a triangle is superimposed), but not during the control condition (discs that change into Pac-Men). We examined this hypothesis by manipulating the perceptual stability of inducers. When the inducers could be perceptually interpreted as persistent circles, we replicated the up- and downregulation pattern shown in previous studies. However, when the inducers could not be perceived as persistent circles, we still observed enhanced activity in neurons representing the illusory shape but the suppression of activity in neurons representing the inducers was absent. Thus our results support the hypothesis that the activity suppression in neurons representing the inducers during the Kanizsa illusion is better explained by neural adaptation to perceptually stable input than by reduced prediction error.


  • Open Access

    Touch engages visual spatial contextual processing

    by Alexis Pérez-Bellido, Ryan D. Pappal, Jeffrey M. Yau

    The spatial context in which we view a visual stimulus strongly determines how we perceive the stimulus. In the visual tilt illusion, the perceived orientation of a visual grating is affected by the orientation signals in its surrounding context. Conceivably, the spatial context in which a visual grating is perceived can be defined by interactive multisensory information rather than visual signals alone. Here, we tested the hypothesis that tactile signals engage the neural mechanisms supporting visual contextual modulation. Because tactile signals also convey orientation information and touch can selectively interact with visual orientation perception, we predicted that tactile signals would modulate the visual tilt illusion. We applied a bias-free method to measure the tilt illusion while testing visual-only, tactile-only or visuo-tactile contextual surrounds. We found that a tactile context can influence visual tilt perception. Moreover, combining visual and tactile orientation information in the surround results in a larger tilt illusion relative to the illusion achieved with the visual-only surround. These results demonstrate that the visual tilt illusion is subject to multisensory influences and imply that non-visual signals access the neural circuits whose computations underlie the contextual modulation of vision.