Biology subjects

Hertz, U.

Publications and source records attributed to Hertz, U..

7 recordsLinked to original sources

bioRxiv2026

Preserved implicit metacognitive sensitivity distinguishes psychosis risk from first-episode psychosis: A cross-sectional virtual reality-based study

Relatively preserved insight distinguishes individuals at risk for psychosis from those with full-blown psychosis. Metacognitive processes thought to support insight and uncertainty monitoring may therefore serve as early markers of illness progression. Yet findings have been inconsistent, perhaps partly due to reliance on explicit confidence ratings that introduce reflection and response biases. To address these limitations, we used a novel implicit confidence measure derived from post-decision gaze in a virtual-reality probabilistic learning task. Gaze-based confidence quantifies the alignment between spatial predictions and gaze direction. We assessed first-order learning and gaze-based metacognition in four groups: clinical high-risk for psychosis (CHR-P), first episode psychosis (FEP), help-seeking controls (HSC), and healthy controls (HC). We tested whether implicit metacognition differentiates psychosis risk from psychosis. Learning accuracy was reduced in both CHR-P and FEP compared to control groups. At the metacognitive level, CHR-P confidence levels were approximately commensurate with their reduced first-order performance, indicating preserved confidence calibration, along with preserved metacognitive sensitivity--the ability to distinguish correct from incorrect decisions. In contrast, FEP showed impaired confidence calibration and reduced metacognitive sensitivity. Metacognitive calibration and sensitivity distinguished CHR-P from FEP and provided predictive value in distinguishing CHR-P from FEP, whereas learning accuracy did not. These findings reveal a dissociation between first-order cognitive processes and distinct aspects of implicit metacognition, including confidence calibration and metacognitive sensitivity, across the psychosis continuum. This dissociation may refine early clinical characterization and improve identification of preserved insight-related mechanisms in psychosis risk.

neuroscience↗

bioRxiv2025

Unreal? A Behavioral, Physiological, and Computational Model of the Sense of Reality

Our awareness of dreams, hallucinations, and illusions reflects an intriguing human capacity to recognize potentially false perceptions, known as the Sense of Reality (SoR). Though central to mental health, the study of SoR has been hindered by the subjective nature of hallucinatory experiences. Here we employed a novel virtual reality paradigm simulating virtual hallucinations, mirroring the phenomenology of clinical hallucinations. Combining psychophysics, physiological recordings, and computational modeling in one exploratory (n = 31) and one preregistered experiment (n = 32), we found that SoR varied with virtual hallucination magnitude and domain. SoR judgments were associated with distinct motor, pupillary, and cardiac responses, allowing classification of virtual hallucination exposure. We present a computational model in which SoR judgments arise from comparing current sensory input to an internal model of the world. Our results shed light on the age-old question how do we know what is real?

neuroscience↗

bioRxiv2024

'Eye Know': Gaze reflects confidence in explicit predictions while relying on a distinct computational mechanism

Decisions are naturally accompanied by a feeling of confidence in their correctness. Typically measured via self-reported judgments, decision confidence is also expressed implicitly through sensorimotor behavior. In one exploratory and two preregistered experiments (total N = 115), we combined VR-based probabilistic-learning and spatial-cueing paradigms with computational modeling to examine whether post-decision gaze can serve as an implicit behavioral measure of decision confidence. Gaze direction generally aligned with predictive decisions about a targets upcoming location, while the degree of alignment tracked the model-based probability underlying the decision. Gaze-decision alignment also exhibited previously established statistical hallmarks of confidence. Under increased uncertainty, gaze and decisions tended to diverge, reflecting strategic "hedging" behavior. Gaze-decision alignment and explicit confidence judgments were modestly correlated, and remained correlated after controlling for the decisions probability, suggesting that the two are partially overlapping manifestations of internal confidence. Notably, reported confidence showed higher metacognitive sensitivity and increased serial dependence across trials, while gaze-based confidence expressed a more trial-specific readout of internal confidence. These findings suggest that post-decision gaze in visual-spatial learning tasks is an implicit, embodied manifestation of confidence. The complementary functional characteristics of implicit and explicit expressions of confidence may support adaptive behavior during learning.

neuroscience↗

bioRxiv2021

Using Deep Learning to Predict Human Decisions and Cognitive Models to Explain Deep Learning Models

Deep neural networks (DNN) models have the potential to provide new insights in the study of human decision making, due to their high capacity and data-driven design. While these models may be able to go beyond theory-driven models in predicting human behaviour, their opaque nature limits their ability to explain how an operation is carried out. This explainability problem remains unresolved. Here we demonstrate the use of a DNN model as an exploratory tool to identify predictable and consistent human behaviour in value-based decision making beyond the scope of theory-driven models. We then propose using theory-driven models to characterise the operation of the DNN model. We trained a DNN model to predict human decisions in a four-armed bandit task. We found that this model was more accurate than a reinforcement-learning reward-oriented model geared towards choosing the most rewarding option. This disparity in accuracy was more pronounced during times when the expected reward from all options was similar, i.e., no unambiguous good option. To investigate this disparity, we introduced a reward-oblivious model, which was trained to predict human decisions without information about the rewards obtained from each option. This model captured decision-sequence patterns made by participants (e.g., a-b-c-d). In a series of experimental offline simulations of all models we found that the general model was in line with a reward-oriented models predictions when one option was clearly better than the others. However, when options expected rewards were similar to each other, it was in-line with the reward-oblivious models pattern completion predictions. These results indicate the contribution of predictable but task-irrelevant decision patterns to human decisions, especially when task-relevant choices are not immediately apparent. Importantly, we demonstrate how theory-driven cognitive models can be used to characterise the operation of DNNs, making them a useful explanatory tool in scientific investigation. Author SummaryDeep neural networks (DNN) models are an extremely useful tool across multiple domains, and specifically for performing tasks that mimic and predict human behaviour. However, due to their opaque nature and high level of complexity, their ability to explain human behaviour is limited. Here we used DNN models to uncover hitherto overlooked aspects of human decision making, i.e., their reliance on predictable patterns for exploration. For this purpose, we trained a DNN model to predict human choices in a decision-making task. We then characterised this data-driven model using explicit, theory-driven cognitive models, in a set of offline experimental simulations. This relationship between explicit and data-driven approaches, where high-capacity models are used to explore beyond the scope of established models and theory-driven models are used to explain and characterise these new grounds, make DNN models a powerful scientific tool.

neuroscience↗

bioRxiv2019

The Neural Basis of Shared Preference Learning

During our daily lives, we often learn about the similarity of the traits and preferences of others to our own and use that information during our social interactions. However, it is unclear how the brain represents similarity between the self and others. One possible mechanism is to track similarity to oneself regardless of the identity of the other (Similarity account); an alternative is to track each confederate in terms of consistency of the similarity to the self, with respect to the choices they have made before (consistency account). Our study combined fMRI and computational modelling of reinforcement learning (RL) to investigate the neural processes that underlie learning about preference similarity. Participants chose which of two pieces of artwork they preferred and saw the choices of one confederate who usually shared their preference and another who usually did not. We modelled neural activation with RL models based on the similarity and consistency accounts. Data showed more brain regions whose activity pattern fits with the consistency account, specifically, areas linked to reward and social cognition. Our findings suggest that impressions of other people can be calculated in a person-specific manner which assumes that each individual behaves consistently with their past choices.

neuroscience↗

bioRxiv2017

Neural Computations Underpinning The Strategic Management Of Influence In Advice Giving

Research on social influence has mainly focused on the target of influence (e.g. consumer, voter), while the cognitive and neurobiological underpinnings of the source of the influence (e.g. marketer, politician) remain unexplored. Here we show that advisers managed their influence over their client strategically, by modulating the confidence of their advice, depending on the interaction between advisers level of influence on the client (i.e., being ignored or chosen by the client), and relative merit (i.e., their accuracy in comparison with a rival). Functional magnetic resonance imaging showed that these sources of social information were tracked in distinct regions of the brains social and valuation systems: relative merit in the medial-prefrontal cortex, and selection by client in the temporo-parietal junction. In addition, increased functional connectivity between these two regions during positive merit trials predicted their effect on behaviour. Both sources of social information modulated the activity in the ventral striatum. These results further our understanding of human interactions and provide a framework for investigating the neurobiology of how we try to influence others.

neuroscience↗

bioRxiv2017

Stochastic satisficing account of choice and confidence in uncertain value-based decisions

Every day we make choices under uncertainty; choosing what route to work or which queue in a supermarket to take, for example. It is unclear how outcome variance, e.g. uncertainty about waiting time in a queue, affects decisions and confidence when outcome is stochastic and continuous. How does one evaluate and choose between an option with unreliable but high expected reward, and an option with more certain but lower expected reward? Here we used an experimental design where two choices payoffs took continuous values, to examine the effect of outcome variance on decision and confidence. We found that our participants probability of choosing the good (high expected reward) option decreased when the good or the bad options payoffs were more variable. Their confidence ratings were affected by outcome variability, but only when choosing the good option. Unlike perceptual detection tasks, confidence ratings correlated only weakly with decisions time, but correlated with the consistency of trial-by-trial choices. Inspired by the satisficing heuristic, we propose a \"stochastic satisficing\" (SSAT) model for evaluating options with continuous uncertain outcomes. In this model, options are evaluated by their probability of exceeding an acceptability threshold, and confidence reports scale with the chosen options thus-defined satisficing probability. Participants decisions were best explained by an expected reward model, while the SSAT model provided the best prediction of decision confidence. We further tested and verified the predictions of this model in a second experiment. Our model and experimental results generalize the models of metacognition from perceptual detection tasks to continuous-value based decisions. Finally, we discuss how the stochastic satisficing account of decision confidence serves psychological and social purposes associated with the evaluation, communication and justification of decision-making.\n\nAuthor SummaryEvery day we make several choices under uncertainty, like choosing a queue in a supermarket. However, the computational mechanisms underlying such decisions remain unknown. For example, how does one choose between an option with unreliable high expected reward, like the volatile express queue, and an option with more certain but lower expected reward in the standard queue? Inspired by bounded rationality and the notion of satisficing, i.e. settling for a good enough option, we propose that such decisions are made by comparing the likelihood of different actions to surpass an acceptability threshold. When facing uncertain decisions, our participants confidence ratings were not consistent with the expected outcomes rewards, but instead followed the satisficing heuristic proposed here. Using an acceptability threshold may be especially useful when evaluating and justifying decisions under uncertainty.

neuroscience↗