Perceptual confidence demonstrates trial-by-trial insight into the precision of audio-visual timing encoding

The precision test of metacognitive sensitivity and confidence criteria

Humans experience feelings of confidence in their decisions. In perception, these feelings are typically accurate - we tend to feel more confident about correct decisions. The degree of insight people have into the accuracy of their decisions is known as metacognitive sensitivity. Currently popular methods of estimating metacognitive sensitivity are subject to interpretive ambiguities because they assume people have normally shaped distributions of different experiences when they are repeatedly exposed to a single input. If this normality assumption is violated, calculations can erroneously underestimate metacognitive sensitivity. Here, we describe a means of estimating metacognitive sensitivity that is more robust to violations of the normality assumption. This improved method can easily be added to standard behavioral experiments, and the authors provide Matlab code to help researchers implement these analyses and experimental procedures.

Independent-channels models of temporal-order judgment revisited: A model comparison

The perception of temporal order or simultaneity of stimuli is almost always explained in terms of independent-channels models, such as perceptual-moment, triggered-moment, and attention-switching models. Independent-channels models generally posit that stimuli are processed in separate peripheral channels and that their arrival-time difference at a central location is translated into an internal state of order (simultaneity) if it reaches (misses) a certain threshold. Non-monotonic and non-parallel psychometric functions in a ternary-response task provided critical evidence against a wide range of independent-channels models. However, two independent-channels models have been introduced in the last decades that can account for such shapes by considering misreports of internal states (response-error model) or by assuming that simultaneity and order judgments rely on distinct sensory and decisional processes (two-stage model). Based on previous ideas, we also consider a two-threshold model, according to which the same arrival-time difference may need to reach a higher threshold for order detection than for successiveness detection. All three models were fitted to various data sets collected over a period of more than a century. The two-threshold model provided the best balance between goodness of fit and parsimony. This preference for the two-threshold model over the two-stage model and the response-error model aligns well with several lines of evidence from cognitive modeling, psychophysics, mental chronometry, and psychophysiology. We conclude that the seemingly deviant shapes of psychometric functions can be explained within the framework of independent-channels models in a simpler way than previously assumed.

Emotional anticipation for dynamic emotional faces is not modulated by schizotypal traits: A Representational Momentum study

Schizotypy, a personality structure that resembles schizophrenia symptoms, is often associated with abnormal facial emotion perception. Based on the prevailing sense of threat in psychotic experiences, and the immediate perceptual history of seeing others' facial expressions, individuals with high schizotypal traits may exhibit a heightened tendency to anticipate anger. To test this, we used insights from Representational Momentum (RM), a perceptual phenomenon in which the endpoint of a dynamic event is systematically displaced forward, into the immediate future. Angry-to-ambiguous and happy-to-ambiguous avatar faces were presented, each followed by a probe with the same (ambiguous) expression as the endpoint, or one slightly changed to express greater happiness/anger. Participants judged if the probe was "equal" to the endpoint and rated how confident they were. The sample was divided into high (N = 46) and low (N = 49) schizotypal traits using the Schizotypal Personality Questionnaire (SPQ). First, a forward bias was found in happy-to-ambiguous faces, suggesting emotional anticipation solely for dynamic faces changing towards a potential threat (anger). This may reflect an adaptative mechanism, as it is safer to anticipate any hostility from a conspecific than the opposite. Second, contrary to our hypothesis, high schizotypal traits did not heighten RM for happy-to-ambiguous faces, nor did they lead to overconfidence in biased judgements. This may suggest a typical pattern of emotional anticipation in non-clinical schizotypy, but caution is needed due to the use of self-report questionnaires, university students, and a modest sample size. Future studies should also investigate if the same holds for clinical manifestations of schizophrenia.

On why we lack confidence in some signal-detection-based analyses of confidence

Signal-detection theory (SDT) is one of the most popular frameworks for analyzing data from studies of human behavior - including investigations of confidence. SDT-based analyses of confidence deliver both standard estimates of sensitivity (d'), and a second estimate informed by high-confidence decisions - meta d'. The extent to which meta d' estimates fall short of d' estimates is regarded as a measure of metacognitive inefficiency, quantifying the contamination of confidence by additional noise. These analyses rely on a key but questionable assumption - that repeated exposures to an input will evoke a normally-shaped distribution of perceptual experiences (the normality assumption). Here we show, via analyses inspired by an experiment and modelling, that when distributions of experience do not conform with the normality assumption, meta d' can be systematically underestimated relative to d'. Our data highlight that SDT-based analyses of confidence do not provide a ground truth measure of human metacognitive inefficiency. We explain why deviance from the normality assumption is especially a problem for some popular SDT-based analyses of confidence, in contrast to other analyses inspired by the SDT framework, which are more robust to violations of the normality assumption.

The implied motion aftereffect changes decisions, but not confidence

Viewing static images depicting movement can result in a motion aftereffect: people tend to categorise direction signals as moving in the opposite direction relative to the implied motion in still photographs. This finding could indicate that inferred motion direction can penetrate sensory processing and change perception. Equally possible, however, is that inferred motion changes decision processes, but not perception. Here we test these two possibilities. Since both categorical decisions and subjective confidence are informed by sensory information, confidence can be informative about whether an aftereffect probably results from changes to perceptual or decision processes. We therefore used subjective confidence as an additional measure of the implied motion aftereffect. In Experiment 1 (implied motion), we find support for decision-level changes only, with no change in subjective confidence. In Experiment 2 (real motion), we find equal changes to decisions and confidence. Our results suggest the implied motion aftereffect produces a bias in decision-making, but leaves perceptual processing unchanged.

An observer model of tilt perception, sensitivity and confidence

Humans experience levels of confidence in perceptual decisions that tend to scale with the precision of their judgements; but not always. Sometimes precision can be held constant while confidence changes-leading researchers to assume precision and confidence are shaped by different types of information (e.g. perceptual and decisional). To assess this, we examined how visual adaptation to oriented inputs changes tilt perception, perceptual sensitivity and confidence. Some adaptors had a greater detrimental impact on measures of confidence than on precision. We could account for this using an observer model, where precision and confidence rely on different magnitudes of sensory information. These data show that differences in perceptual sensitivity and confidence can therefore emerge, not because these factors rely on different types of information, but because they rely on different magnitudes of sensory information.

Neural correlates of subjective timing precision and confidence

Humans perceptual judgments are imprecise, as repeated exposures to the same physical stimulation (e.g. audio-visual inputs separated by a constant temporal offset) can result in different decisions. Moreover, there can be marked individual differences – precise judges will repeatedly make the same decision about a given input, whereas imprecise judges will make different decisions. The causes are unclear. We examined this using audio-visual (AV) timing and confidence judgments, in conjunction with electroencephalography (EEG) and multivariate pattern classification analyses. One plausible cause of differences in timing precision is that it scales with variance in the dynamics of evoked brain activity. Another possibility is that equally reliable patterns of brain activity are evoked, but there are systematic differences that scale with precision. Trial-by-trial decoding of input timings from brain activity suggested precision differences may not result from variable dynamics. Instead, precision was associated with evoked responses that were exaggerated (more different from baseline) ~300 ms after initial physical stimulations. We suggest excitatory and inhibitory interactions within a winner-take-all neural code for AV timing might exaggerate responses, such that evoked response magnitudes post-stimulation scale with encoding success.

Predictable events elicit less visual and temporal information uptake in an oddball paradigm

In the visual oddball paradigm, surprising inputs can seem expanded in time relative to unsurprising repeated events. A horizontal input embedded in a train of successive vertical inputs can, for instance, seem relatively protracted in time, even if all inputs are presented for an identical duration. It is unclear if this effect results from surprising events becoming apparently protracted, or from repeated events becoming apparently contracted in time. To disambiguate, we used a non-relative duration reproduction task, in which several standards preceded a test stimulus that had to be reproduced. We manipulated the predictability of test content over successive presentations. Overall, our data suggest that predictable stimuli induce a contraction of apparent duration (Experiments 1, 3, and 4). We also examine sensitivity to test content, and find that predictable stimuli elicit less uptake of visual information (Experiments 2 and 3). We discuss these findings in relation to the predictive coding framework.