Biology subjects

Haarsma, J.

Publications and source records attributed to Haarsma, J..

10 recordsLinked to original sources

bioRxiv2026

The post-hoc montage of perception: deep layers of primary visual cortex encode postdictive percepts

One of the most puzzling aspects of perception is postdiction: later information can change how a previous stimulus is perceived [1,2]. This indicates that perception is not a livestream of the external world, but rather a post-hoc reconstruction. How does the brain revise perception? It has been proposed that recurrent processing in the sensory hierarchy may sustain earlier information long enough for later information to impact it [3]. We hypothesised that this allows feedback from higher-order multisensory regions to alter the representation of earlier visual stimuli in the primary visual cortex (V1) [4]. Here, we tested this hypothesis using two postdictive illusions [5], where the presence or absence of a tone determines whether an illusory flash is perceived, or an actually presented flash is perceptually suppressed, respectively. This allowed us to isolate neural signals reflecting illusory percepts from sensory inputs in retinotopic space. We collected 7T layer-specific fMRI data while participants (N=23) experienced the illusions. We found that postdictive illusions induced similar activity patterns as real percepts, specifically in the deep layers of V1, suggesting postdictive feedback. Informational connectivity analyses suggested that the source of this feedback was the Superior Temporal Cortex (STC). We conclude that the postdictive montage of perception is reflected in V1, implemented through a delaying feedback loop with higher-order sensory cortex. This extends the role V1 plays in perceptual inference: beyond receiving sensory inputs reflecting the present and predictions of the future, it is involved in the post-hoc revision of perception.

neuroscience↗

bioRxiv2024

Pre-stimulus alpha oscillations encode stimulus-specific visual predictions

Predictions of future events have a major impact on how we process sensory signals. However, it remains unclear how the brain keeps predictions online in anticipation of future inputs. Here, we combined magnetoencephalography (MEG) and multivariate decoding techniques to investigate the content of perceptual predictions and their frequency characteristics. Participants were engaged in a shape discrimination task, while auditory cues predicted which specific shape would likely appear. Frequency analysis revealed significant oscillatory fluctuations of predicted shape representations in the pre-stimulus window in the alpha band (10 - 11Hz). Furthermore, we found that this stimulus-specific alpha power was linked to expectation effects on shape discrimination. Our findings demonstrate that sensory predictions are embedded in pre-stimulus alpha oscillations and modulate subsequent perceptual performance, providing a neural mechanism through which the brain deploys perceptual predictions.

neuroscience↗

bioRxiv2024

Expectations about presence enhance the influence of content-specific expectations on low-level orientation judgements

Will something appear and if so, what will it be? Perceptual expectations can concern both the presence and content of a stimulus. However, it is unclear how these different types of expectations interact with each other in biasing perception. Here, we tested how expectations about stimulus presence and content differently affect perceptual inference. Across separate online discovery (N=110) and replication samples (N=218), participants were asked to judge both the presence and content (orientation) of noisy grating stimuli. Crucially, preceding compound cues simultaneously and orthogonally predicted both whether a grating was likely to appear as well as what its orientation would be. We found that expectations of presence interacted with expectations of content, such that the latters effect on discrimination was larger when a stimulus was expected to appear than when it was not. This interaction was observed both when a grating was truly presented and when participants falsely perceived one. Confidence in having seen a grating on the other hand was independently affected by presence and content expectations. Further, modelling revealed higher sensitivity in distinguishing between grating presence and absence following absence cues than presence cues, demonstrating an asymmetry between gathering evidence in favour of stimulus presence and absence. Finally, evidence for overweighted predictions being associated with hallucination-like perception was inconclusive. In sum, our results provide nuance to popular predictive processing accounts of perception by showing that expectations of presence and content have distinct but interacting roles in shaping conscious perception.

neuroscience↗

bioRxiv2023

Shared and diverging neural dynamics underlying false and veridical perception

We often mistake visual noise for meaningful images, which sometimes appear to be as convincing as veridical percepts. This suggests that there is considerable overlap between the mechanisms that underlie false and veridical perception. Yet, false percepts must arise at least in part from internally generated signals. Here, we apply multivariate analyses to human MEG data to study the overlap between veridical and false perception across two discrete stages of perceptual inference: discrimination of content (what did I see) and detection (did I see something?). To this end, participants performed a visual discrimination task requiring them to indicate the orientation of a noisy grating, as well as their confidence in having seen a grating. Importantly, on 50% of trials no gratings were presented (noise-only trials). On a subset of these noise-only trials, participants reported seeing a grating with high confidence, dubbed here false percepts. We found that a sensory signal reflecting the content of these false percepts was present both before and after stimulus onset. Uniquely, high confidence false, but not veridical, percepts were associated with increased pre-stimulus high alpha/low beta [11-14Hz] power, potentially reflecting enhanced reliance on top-down signalling on false percept trials. Later on, a shared neural code reflecting confidence in stimulus presence emerged for both false and veridical percepts, as revealed by cross-decoding. These findings suggest that false percepts arise through sensory-like signals reflecting both content and detection signals, similar to veridical percepts, with an increase in pre-stimulus alpha/beta power uniquely contributing to false percepts. Significance statementThe neural mechanisms underlying false percepts are likely different from those that underlie veridical perception, as the former are generated endogenously, whereas the latter are the result of an external stimulus. Yet, false percepts often get confused for veridical perception, suggesting a converging mechanism. This study explores the extent to which the mechanisms diverge and converge. We found that both high confidence false and veridical percepts were accompanied by content-specific stimulus-like orientation signals, as well as a shared signal reflecting perceptual confidence. In contrast, we found that false, but not veridical, percepts were preceded by increased high alpha/low beta [11-14 Hz] power, possibly reflecting a reliance on endogenous signals.

neuroscience↗

bioRxiv2023

Predictions and errors are distinctly represented across V1 layers

Predictive processing frameworks of cortical functioning propose that neural populations in different cortical layers serve distinct roles in representing the world. There are distinct testable theories within this framework that we examined with a 7T fMRI study, where we contrasted responses in primary visual cortex (V1) to expected (75% likely) and unexpected (25%) Gabor orientations. Multivariate decoding analyses revealed an interaction between expectation and layer, such that expected events could be decoded with comparable accuracy across layers, while unexpected events could only be decoded in superficial laminae. These results are in line with predictive processing accounts where expected virtual input is injected into deep layers, while superficial layers process the error with respect to expected signals. While this account of cortical processing has been popular for decades, such distinctions have not previously been demonstrated in the human sensory brain. We discuss how both prediction and error processes may operate together to shape our unitary perceptual experiences.

neuroscience↗

bioRxiv2022

Perceptual expectations and false percepts generate stimulus-specific activity in distinct layers of the early visual cortex

Perception has been proposed to result from the integration of feedforward sensory signals with internally generated feedback signals. The latter are believed to play an important role in driving false percepts, i.e., seeing things that are not actually there. Feedforward and feedback influences on perception can be studied using layer-specific fMRI, which we used here to interrogate neural activity underlying high confidence false percepts while healthy participants (N=25) performed a perceptual orientation discrimination task. Orientation-specific BOLD activity in the deep and superficial layers of V2 reflected perceptual expectations induced by predictive auditory cues. However, these expectations did not influence participants perception. Instead, high confidence false percepts were reflected by orientation-specific activity in the middle input layers of V2, suggesting a feedforward signal contributing to false percepts. The prevalence of high confidence false percepts was related to everyday hallucination severity in a separate online sample (N=100), suggesting a possible link with abnormal perceptual experiences. These results reveal a feedforward mechanism underlying false percepts, reflected by spontaneous stimulus-like activity in the input layers of the visual cortex, independent of top-down perceptual expectations.

neuroscience↗

bioRxiv2019

Reinforcement learning as an intermediate phenotype in psychosis? Deficits sensitive to illness stage but not associated with polygenic risk of schizophrenia in the general population

BackgroundSchizophrenia is a complex disorder in which the causal relations between risk genes and observed clinical symptoms are not well understood and the explanatory gap is too wide to be clarified without considering an intermediary level. Thus, we aimed to test the hypothesis of a pathway from molecular polygenic influence to clinical presentation occurring via deficits in reinforcement learning. MethodsWe administered a reinforcement learning task (Go/NoGo) that measures reinforcement learning and the effect of Pavlovian bias on decision making. We modelled the behavioural data with a hierarchical Bayesian approach (hBayesDM) to decompose task performance into its underlying learning mechanisms. Study 1 included controls (n= 29, F|M=0.81), At Risk Mental State for psychosis (ARMS, n= 23, F|M=0.35) and FEP (First-episode psychosis, n= 26, F|M=0.18). Study 2 included healthy adolescents (n= 735, F|M= 1.06), 390 of whom had their polygenic risk scores for schizophrenia (PRSs) calculated. ResultsPatients with FEP showed significant impairments in overriding Pavlovian conflict, a lower learning rate and a lower sensitivity to both reward and punishment. Less widespread deficits were observed in ARMS. PRSs did not significantly predict performance on the task in the general population, which only partially correlated with measures of psychopathology. ConclusionsReinforcement learning deficits are observed in first episode psychosis and, to some extent, in those at clinical risk for psychosis, and were not predicted by molecular genetic risk for schizophrenia in healthy individuals. The study does not support the role of reinforcement learning as an intermediate phenotype in psychosis.

neuroscience↗

bioRxiv2019

Precision-weighting of superior frontal cortex unsigned prediction error signals benefits learning, is mediated by dopamine, and is impaired in psychosis.

Recent theories of cortical function construe the brain as performing hierarchical Bayesian inference. According to these theories, the precision of cortical unsigned prediction error (i.e., surprise) signals plays a key role in learning and decision-making, to be controlled by dopamine, and to contribute to the pathogenesis of psychosis. To test these hypotheses, we studied learning with variable outcome-precision in healthy individuals after dopaminergic modulation and in patients with early psychosis. Behavioural computational modelling indicated that precision-weighting of unsigned prediction errors benefits learning in health, and is impaired in psychosis. FMRI revealed coding of unsigned prediction errors relative to their precision in bilateral superior frontal gyri and dorsal anterior cingulate, which was perturbed by dopaminergic modulation, impaired in psychosis, and associated with task performance and schizotypy. We conclude that precision-weighting of cortical prediction error signals is a key mechanism through which dopamine modulates inference and contributes to the pathogenesis of psychosis.

neuroscience↗

bioRxiv2018

Influence of prior beliefs on perception in early psychosis: effects of illness stage and hierarchical level of belief

Alterations in the balance between prior expectations and sensory evidence may account for faulty perceptions and inferences leading to psychosis. However, uncertainties remain about the nature of altered prior expectations and the degree to which they vary with the emergence of psychosis. We explored how expectations arising at two different levels - cognitive and perceptual - influenced processing of sensory information and whether relative influences of higher and lower level priors differed across people with prodromal symptoms and those with psychotic illness. In two complementary auditory perception experiments, 91 participants (30 with first episode psychosis, 29 at clinical risk for psychosis, and 32 controls) were required to decipher a phoneme within ambiguous auditory input. Expectations were generated in two ways: an accompanying visual input of lip movements observed during auditory presentation, or through written presentation of a phoneme provided prior to auditory presentation. We determined how these different types of information shaped auditory perceptual experience, how this was altered across the prodromal and established phases of psychosis, and how this relates to cingulate glutamate levels assessed by magnetic resonance spectroscopy. The psychosis group relied more on high level cognitive priors compared to both healthy controls and those at clinical risk for psychosis, and more on low level perceptual priors than the clinical risk group. The risk group were marginally less reliant on low level perceptual priors than controls. The results are consistent with previous theory that influences of prior expectations in psychosis in perception differ according to level of prior and illness phase.\n\nGeneral scientific summaryWhat we perceive and believe on any given moment will allow us to form expectations about what we will experience in the next. In psychosis, it is believed that the influence of these so-called perceptual and cognitive prior expectations on perception is altered, thereby giving rise to the symptoms seen in psychosis. However, research thus far has found mixed evidence, some suggesting an increase in the influence of priors and some finding a decrease. Here we test the hypothesis that perceptual and cognitive priors are differentially affected in individuals at-risk for psychosis and individuals with a first episode of psychosis, thereby partially explaining the mixed findings in the literature. We indeed found evidence in favour of this hypothesis, finding weaker perceptual priors in individuals at-risk, but stronger cognitive priors in individuals with first episode psychosis.

neuroscience↗

bioRxiv2018

Precision weighting of cortical unsigned prediction errors is mediated by dopamine and benefits learning

The predictive coding framework construes the brain as performing a specific form of hierarchical Bayesian inference. In this framework the precision of cortical unsigned prediction error (surprise) signals is proposed to play a key role in learning and decision-making, and to be controlled by dopamine. To test this hypothesis, we re-analysed an existing data-set from healthy individuals who received a dopamine agonist, antagonist or placebo and who performed an associative learning task under different levels of outcome precision. Computational reinforcement-learning modelling of behaviour provided support for precision-weighting of unsigned prediction errors. Functional MRI revealed coding of unsigned prediction errors relative to their precision in bilateral superior frontal gyri and dorsal anterior cingulate. Cortical precision-weighting was (i) perturbed by the dopamine antagonist sulpiride, and (ii) associated with task performance. These findings have important implications for understanding the role of dopamine in reinforcement learning and predictive coding in health and illness.

neuroscience↗