Confidence reflects a noisy decision reliability estimate

Fork in the road: How self-efficacy related to walking across terrain influences gaze behavior and path choice

Decisions about where to move occur throughout the day and are essential to life. Different movements may present different challenges and affect the likelihood of achieving a goal. Certain choices may have unintended consequences, some of which may cause harm and bias the decision. Movement decisions rely on a person gathering necessary visual information via shifts in gaze. Here we sought to understand what influences this information-seeking gaze behavior. Participants chose between walking across one of two paths that consisted of terrain images found in either hiking or urban environments. We manipulated the number and type of terrain of each path, which altered the amount of available visual information. We recorded gaze behavior during the approach to the paths and had participants rate the confidence in their ability to walk across each terrain type (i.e., self-efficacy) as though it was real. Participants did not direct gaze more to the path with greater visual information, regardless of how we quantified information. Rather, we show that a person's perception of their motor abilities predicts how they visually explore the environment with their eyes as well as their choice of action. The greater the self-efficacy in walking across one path, the more they directed gaze to it and the more likely they chose to walk across it.

How does V1 population activity inform perceptual certainty?

Neural population activity in sensory cortex informs our perceptual interpretation of the environment. Oftentimes, this population activity will support multiple alternative interpretations. The larger the spread of probability over different alternatives, the more uncertain the selected perceptual interpretation. We test the hypothesis that the reliability of perceptual interpretations can be revealed through simple transformations of sensory population activity. We recorded V1 population activity in fixating macaques while presenting oriented stimuli under different levels of nuisance variability and signal strength. We developed a decoding procedure to infer from V1 activity the most likely stimulus orientation as well as the certainty of this estimate. Our analysis shows that response magnitude, response dispersion, and variability in response gain all offer useful proxies for orientation certainty. Of these three metrics, the last one has the strongest association with the decoder's uncertainty estimates. These results clarify that the nature of neural population activity in sensory cortex provides downstream circuits with multiple options to assess the reliability of perceptual interpretations.

Abstract deliberation by visuomotor neurons in prefrontal cortex

During visually guided behavior, the prefrontal cortex plays a pivotal role in mapping sensory inputs onto appropriate motor plans. When the sensory input is ambiguous, this involves deliberation. It is not known whether the deliberation is implemented as a competition between possible stimulus interpretations or between possible motor plans. Here we study neural population activity in the prefrontal cortex of macaque monkeys trained to flexibly report perceptual judgments of ambiguous visual stimuli. We find that the population activity initially represents the formation of a perceptual choice before transitioning into the representation of the motor plan. Stimulus strength and prior expectations both bear on the formation of the perceptual choice, but not on the formation of the action plan. These results suggest that prefrontal circuits involved in action selection are also used for the deliberation of abstract propositions divorced from a specific motor plan, thus providing a crucial mechanism for abstract reasoning.

Knowing that you know that you know? An extreme-confidence heuristic can lead to above-chance discrimination of metacognitive performance

In daily life, we can not only estimate confidence in our inferences ('I'm sure I failed that exam'), but can also estimate whether those feelings of confidence are good predictors of decision accuracy ('I feel sure I failed, but my feeling is probably wrong; I probably passed'). In the lab, by using simple perceptual tasks and collecting trial-by-trial confidence ratings visual metacognition research has repeatedly shown that participants can successfully predict the accuracy of their perceptual choices. Can participants also successfully evaluate 'confidence in confidence' in these tasks? This is the question addressed in this study. Participants performed a simple, two-interval forced choice numerosity task framed as an exam. Confidence judgements were collected in the form of a 'predicted exam grade'. Finally, we collected 'meta-metacognitive' reports in a two-interval forced-choice design: trials were presented in pairs, and participants had to select that in which they thought their confidence (predicted grade) best matched their accuracy (actual grade), effectively minimizing their quadratic scoring rule (QSR) score. Participants successfully selected trials on which their metacognition was better when metacognitive performance was quantified using area under the type 2 ROC (AUROC2) but not when using the 'gold-standard' measure m-ratio. However, further analyses suggested that participants selected trials on which AUROC2 is lower in part via an extreme-confidence heuristic, rather than through explicit evaluation of metacognitive inferences: when restricting analyses to trials on which participants gave the same confidence rating AUROC2 no longer differed as a function of selection, and likewise when we excluded trials on which extreme confidence ratings were given. Together, our results show that participants are able to make effective metacognitive discriminations on their visual confidence ratings, but that explicit 'meta-metacognitive' processes may not be required.

Distinct neural markers of evidence accumulation index metacognitive processing before and after simple visual decisions

Perceptual decision-making is affected by uncertainty arising from the reliability of incoming sensory evidence (perceptual uncertainty) and the categorization of that evidence relative to a choice boundary (categorical uncertainty). Here, we investigated how these factors impact the temporal dynamics of evidence processing during decision-making and subsequent metacognitive judgments. Participants performed a motion discrimination task while electroencephalography was recorded. We manipulated perceptual uncertainty by varying motion coherence, and categorical uncertainty by varying the angular offset of motion signals relative to a criterion. After each trial, participants rated their desire to change their mind. High uncertainty impaired perceptual and metacognitive judgments and reduced the amplitude of the centro-parietal positivity, a neural marker of evidence accumulation. Coherence and offset affected the centro-parietal positivity at different time points, suggesting that perceptual and categorical uncertainty affect decision-making in sequential stages. Moreover, the centro-parietal positivity predicted participants' metacognitive judgments: larger predecisional centro-parietal positivity amplitude was associated with less desire to change one's mind, whereas larger postdecisional centro-parietal positivity amplitude was associated with greater desire to change one's mind, but only following errors. These findings reveal a dissociation between predecisional and postdecisional evidence processing, suggesting that the CPP tracks potentially distinct cognitive processes before and after a decision.

Confidence ratings do not distinguish imagination from reality

Perceptual reality monitoring refers to the ability to distinguish internally triggered imagination from externally triggered reality. Such monitoring can take place at perceptual or cognitive levels-for example, in lucid dreaming, perceptual experience feels real but is accompanied by a cognitive insight that it is not real. We recently developed a paradigm to reveal perceptual reality monitoring errors during wakefulness in the general population, showing that imagined signals can be erroneously attributed to perception during a perceptual detection task. In the current study, we set out to investigate whether people have insight into perceptual reality monitoring errors by additionally measuring perceptual confidence. We used hierarchical Bayesian modeling of confidence criteria to characterize metacognitive insight into the effects of imagery on detection. Over two experiments, we found that confidence criteria moved in tandem with the decision criterion shift, indicating a failure of reality monitoring not only at a perceptual but also at a metacognitive level. These results further show that such failures have a perceptual rather than a decisional origin. Interestingly, offline queries at the end of the experiment revealed global, task-level insight, which was uncorrelated with local, trial-level insight as measured with confidence ratings. Taken together, our results demonstrate that confidence ratings do not distinguish imagination from reality during perceptual detection. Future research should further explore the different cognitive dimensions of insight into reality judgments and how they are related.

Response sub-additivity and variability quenching in visual cortex

Sub-additivity and variability are ubiquitous response motifs in the primary visual cortex (V1). Response sub-additivity enables the construction of useful interpretations of the visual environment, whereas response variability indicates the factors that limit the precision with which the brain can do this. There is increasing evidence that experimental manipulations that elicit response sub-additivity often also quench response variability. Here, we provide an overview of these phenomena and suggest that they may have common origins. We discuss empirical findings and recent model-based insights into the functional operations, computational objectives and circuit mechanisms underlying V1 activity. These different modelling approaches all predict that response sub-additivity and variability quenching often co-occur. The phenomenology of these two response motifs, as well as many of the insights obtained about them in V1, generalize to other cortical areas. Thus, the connection between response sub-additivity and variability quenching may be a canonical motif across the cortex.

Foraging Under Uncertainty Follows the Marginal Value Theorem with Bayesian Updating of Environment Representations

Foraging theory has been a remarkably successful approach to understanding the behavior of animals in many contexts. In patch-based foraging contexts, the marginal value theorem (MVT) shows that the optimal strategy is to leave a patch when the marginal rate of return declines to the average for the environment. However, the MVT is only valid in deterministic environments whose statistics are known to the forager; naturalistic environments seldom meet these strict requirements. As a result, the strategies used by foragers in naturalistic environments must be empirically investigated. We developed a novel behavioral task and a corresponding computational framework for studying patch-leaving decisions in head-fixed and freely moving mice. We varied between-patch travel time, as well as within-patch reward depletion rate, both deterministically and stochastically. We found that mice adopt patch residence times in a manner consistent with the MVT and not explainable by simple ethologically motivated heuristic strategies. Critically, behavior was best accounted for by a modified form of the MVT wherein environment representations were updated based on local variations in reward timing, captured by a Bayesian estimator and dynamic prior. Thus, we show that mice can strategically attend to, learn from, and exploit task structure on multiple timescales simultaneously, thereby efficiently foraging in volatile environments. The results provide a foundation for applying the systems neuroscience toolkit in freely moving and head-fixed mice to understand the neural basis of foraging under uncertainty.

What's in a sample? Epistemic uncertainty and metacognitive awareness in risk taking

In a fundamentally uncertain world, sound information processing is a prerequisite for effective behavior. Given that information processing is subject to inevitable cognitive imprecision, decision makers should adapt to this imprecision and to the resulting epistemic uncertainty when taking risks. We tested this metacognitive ability in two experiments in which participants estimated the expected value of different number distributions from sequential samples and then bet on their own estimation accuracy. Results show that estimates were imprecise, and this imprecision increased with higher distributional standard deviations. Importantly, participants adapted their risk-taking behavior to this imprecision and hence deviated from the predictions of Bayesian models of uncertainty that assume perfect integration of information. To explain these results, we developed a computational model that combines Bayesian updating with a metacognitive awareness of cognitive imprecision in the integration of information. Modeling results were robust to the inclusion of an empirical measure of participants' perceived variability. In sum, we show that cognitive imprecision is crucial to understanding risk taking in decisions from experience. The results further demonstrate the importance of metacognitive awareness as a cognitive building block for adaptive behavior under (partial) uncertainty.

Seeing in crowds: Averaging first, then max

Crowding, a fundamental limit in object recognition, is believed to result from excessive integration of nearby items in peripheral vision. To understand its pooling mechanisms, we measured subjects' internal response distributions in an orientation crowding task. Contrary to the prediction of an averaging model, we observed a pattern suggesting that the perceptual judgement is made based on choosing the largest response across the noise-perturbed items. A model featuring first-stage averaging and second-stage signed-max operation predicts the diverse errors made by human observers under various signal strength levels. These findings suggest that different rules operate to resolve the bottleneck at early and high-level stages of visual processing, implementing a combination of linear and nonlinear pooling strategies.

Subjective Confidence as a Monitor of the Replicability of the Response

Confidence is commonly assumed to monitor the accuracy of responses. However, intriguing results, examined in the light of philosophical discussions of epistemic justification, suggest that confidence actually monitors the reliability of choices rather than (directly) their accuracy. The focus on reliability is consistent with the view that the construction of truth has much in common with the construction of reality: extracting reliable properties that afford prediction. People are assumed to make a binary choice by sampling cues from a "collective wisdomware," and their confidence is based on the consistency of these cues, in line with the self-consistency model. Here, however, I propose that internal consistency is taken to index the reliability of choices themselves-the likelihood that they will be repeated. The results of 10 studies using binary decisions from different domains indicated that confidence in a choice predicts its replicability both within individuals and across individuals. This was so for domains for which choices have a truth value and for those for which they do not. For the former domains, differences in replicability mediated the prediction of accuracy whether confidence was diagnostic or counterdiagnostic of accuracy. Metatheoretical, methodological, and practical implications are discussed.

Human-like dissociations between confidence and accuracy in convolutional neural networks

Prior research has shown that manipulating stimulus energy by changing both stimulus contrast and variability results in confidence-accuracy dissociations in humans. Specifically, even when performance is matched, higher stimulus energy leads to higher confidence. The most common explanation for this effect is the positive evidence heuristic where confidence neglects evidence that disconfirms the choice. However, an alternative explanation is the signal-and-variance-increase hypothesis, according to which these dissociations arise from low-level changes in the separation and variance of perceptual representations. Because artificial neural networks lack built-in confidence heuristics, they can serve as a test for the necessity of confidence heuristics in explaining confidence-accuracy dissociations. Therefore, we tested whether confidence-accuracy dissociations induced by stimulus energy manipulations emerge naturally in convolutional neural networks (CNNs). We found that, across three different energy manipulations, CNNs produced confidence-accuracy dissociations similar to those found in humans. This effect was present for a range of CNN architectures from shallow 4-layer networks to very deep ones, such as VGG-19 and ResNet -50 pretrained on ImageNet. Further, we traced back the reason for the confidence-accuracy dissociations in all CNNs to the same signal-and-variance increase that has been proposed for humans: higher stimulus energy increased the separation and variance of the CNNs' internal representations leading to higher confidence even for matched accuracy. These findings cast doubt on the necessity of the positive evidence heuristic to explain human confidence and establish CNNs as promising models for adjudicating between low-level, stimulus-driven and high-level, cognitive explanations of human behavior.

Metacognition and Confidence: A Review and Synthesis

Determining the psychological, computational, and neural bases of confidence and uncertainty holds promise for understanding foundational aspects of human metacognition. While a neuroscience of confidence has focused on the mechanisms underpinning subpersonal phenomena such as representations of uncertainty in the visual or motor system, metacognition research has been concerned with personal-level beliefs and knowledge about self-performance. I provide a road map for bridging this divide by focusing on a particular class of confidence computation: propositional confidence in one's own (hypothetical) decisions or actions. Propositional confidence is informed by the observer's models of the world and their cognitive system, which may be more or less accurate-thus explaining why metacognitive judgments are inferential and sometimes diverge from task performance. Disparate findings on the neural basis of uncertainty and performance monitoring are integrated into a common framework, and a new understanding of the locus of action of metacognitive interventions is developed.

Introspective inference counteracts perceptual distortion

Introspective agents can recognize the extent to which their internal perceptual experiences deviate from the actual states of the external world. This ability, also known as insight, is critically required for reality testing and is impaired in psychosis, yet little is known about its cognitive underpinnings. We develop a Bayesian modeling framework and a psychophysics paradigm to quantitatively characterize this type of insight while people experience a motion after-effect illusion. People can incorporate knowledge about the illusion into their decisions when judging the actual direction of a motion stimulus, compensating for the illusion (and often overcompensating). Furthermore, confidence, reaction-time, and pupil-dilation data all show signatures consistent with inferential adjustments in the Bayesian insight model. Our results suggest that people can question the veracity of what they see by making insightful inferences that incorporate introspective knowledge about internal distortions.

Modality independent or modality specific? Common computations underlie confidence judgements in visual and auditory decisions

The mechanisms that enable humans to evaluate their confidence across a range of different decisions remain poorly understood. To bridge this gap in understanding, we used computational modelling to investigate the processes that underlie confidence judgements for perceptual decisions and the extent to which these computations are the same in the visual and auditory modalities. Participants completed two versions of a categorisation task with visual or auditory stimuli and made confidence judgements about their category decisions. In each modality, we varied both evidence strength, (i.e., the strength of the evidence for a particular category) and sensory uncertainty (i.e., the intensity of the sensory signal). We evaluated several classes of computational models which formalise the mapping of evidence strength and sensory uncertainty to confidence in different ways: 1) unscaled evidence strength models, 2) scaled evidence strength models, and 3) Bayesian models. Our model comparison results showed that across tasks and modalities, participants take evidence strength and sensory uncertainty into account in a way that is consistent with the scaled evidence strength class. Notably, the Bayesian class provided a relatively poor account of the data across modalities, particularly in the more complex categorisation task. Our findings suggest that a common process is used for evaluating confidence in perceptual decisions across domains, but that the parameter settings governing the process are tuned differently in each modality. Overall, our results highlight the impact of sensory uncertainty on confidence and the unity of metacognitive processing across sensory modalities.