Measuring mislocalization of angle vertices.

Localisation of simple stimuli such as angle vertices may contribute to a plethora of illusory effects. We focus on the Müller-Lyer illusion in an attempt to measure and characterise a more elementary effect that may contribute to the magnitude of said illusion. Perceived location error of angle vertices (a single set of Müller-Lyer fins) and arcs in a 2D plane was measured with the aim to provide clarification of ambiguous results from studies of angle localisation and expand the results to other types of stimuli. In three experiments, we utilised the method of constant stimuli in order to determine perceived locations of angle vertices (Experiments 1 and 2) as well as circular and elliptical arcs (Experiment 3). The results show significant distortions of perceived compared to objective vertex locations (all effect sizes d > 1.01, p < .001). Experiment 2 revealed strong effects of angle size and fin length on localisation error. Mislocalization was larger for more acute angles and longer angle fins (both ηp² = .43, p < .001). In Experiment 3, localisation errors were larger for longer arcs (ηp² = .19, p = .001) irrespective of shape (circular or elliptical). We discuss the effect in the context of modern trends in research of the Müller-Lyer illusion as well as the widely popular centroid theory. (PsycInfo Database Record (c) 2024 APA, all rights reserved).

Clarifying status of DNNs as models of human vision.

On several key issues we agree with the commentators. Perhaps most importantly, everyone seems to agree that psychology has an important role to play in building better models of human vision, and (most) everyone agrees (including us) that deep neural networks (DNNs) will play an important role in modelling human vision going forward. But there are also disagreements about what models are for, how DNN-human correspondences should be evaluated, the value of alternative modelling approaches, and impact of marketing hype in the literature. In our view, these latter issues are contributing to many unjustified claims regarding DNN-human correspondences in vision and other domains of cognition. We explore all these issues in this response.

Human shape representations are not an emergent property of learning to classify objects.

Humans are particularly sensitive to relationships between parts of objects. It remains unclear why this is. One hypothesis is that relational features are highly diagnostic of object categories and emerge as a result of learning to classify objects. We tested this by analyzing the internal representations of supervised convolutional neural networks (CNNs) trained to classify large sets of objects. We found that CNNs do not show the same sensitivity to relational changes as previously observed for human participants. Furthermore, when we precisely controlled the deformations to objects, human behavior was best predicted by the number of relational changes while CNNs were equally sensitive to all changes. Even changing the statistics of the learning environment by making relations uniquely diagnostic did not make networks more sensitive to relations in general. Our results show that learning to classify objects is not sufficient for the emergence of human shape representations. Instead, these results suggest that humans are selectively sensitive to relational changes because they build representations of distal objects from their retinal images and interpret relational changes as changes to these distal objects. This inferential process makes human shape representations qualitatively different from those of artificial neural networks optimized to perform image classification. (PsycInfo Database Record (c) 2023 APA, all rights reserved).

Single-Heuristic Reasoning: Is It Still Dual-Process?

Traditionally, paradigms used to study conflict in reasoning (and metacognition during reasoning) pit heuristic processes against analytical processes. Findings indicate that the presence of conflict between processes prolongs reasoning and decreases accuracy and confidence. In this study, we aimed to explore reasoning and metacognition when only one type of heuristic process is exploited to cue multiple responses. In two experiments, a novel modification of the Base Rate neglect task was used to create versions in which one belief-based heuristic competes, or works in concert, with another of the same type to provide a response. Experiment 1 results reveal that the presence of conflict between cued responses does not affect meta-reasoning, which indicates that reasoning defaulted to a single process. An alternative explanation was that the effect of conflict was masked due to an imbalance in the strength of the dominant response between conflicting and congruent versions. Experiment 2 was designed to test hypotheses based on these competing explanations. Findings show that when the strength of a response was no longer masking the effect, the conflict did result in longer reasoning times and lower confidence. The study provides more robust evidence in favor of the dual-process account of reasoning, introduces a new methodological approach, and discusses how conflict may be modulated during reasoning.

Deep problems with neural network models of human vision.

Deep neural networks (DNNs) have had extraordinary successes in classifying photographic images of objects and are often described as the best models of biological vision. This conclusion is largely based on three sets of findings: (1) DNNs are more accurate than any other model in classifying images taken from various datasets, (2) DNNs do the best job in predicting the pattern of human errors in classifying objects taken from various behavioral datasets, and (3) DNNs do the best job in predicting brain signals in response to images taken from various brain datasets (e.g., single cell responses or fMRI data). However, these behavioral and brain datasets do not test hypotheses regarding what features are contributing to good predictions and we show that the predictions may be mediated by DNNs that share little overlap with biological vision. More problematically, we show that DNNs account for almost no results from psychological research. This contradicts the common claim that DNNs are good, let alone the best, models of human object recognition. We argue that theorists interested in developing biologically plausible models of human vision need to direct their attention to explaining psychological findings. More generally, theorists need to build models that explain the results of experiments that manipulate independent variables designed to test hypotheses rather than compete on making the best predictions. We conclude by briefly summarizing various promising modeling approaches that focus on psychological data.

Feature blindness: A challenge for understanding and modelling visual object recognition.

Humans rely heavily on the shape of objects to recognise them. Recently, it has been argued that Convolutional Neural Networks (CNNs) can also show a shape-bias, provided their learning environment contains this bias. This has led to the proposal that CNNs provide good mechanistic models of shape-bias and, more generally, human visual processing. However, it is also possible that humans and CNNs show a shape-bias for very different reasons, namely, shape-bias in humans may be a consequence of architectural and cognitive constraints whereas CNNs show a shape-bias as a consequence of learning the statistics of the environment. We investigated this question by exploring shape-bias in humans and CNNs when they learn in a novel environment. We observed that, in this new environment, humans (i) focused on shape and overlooked many non-shape features, even when non-shape features were more diagnostic, (ii) learned based on only one out of multiple predictive features, and (iii) failed to learn when global features, such as shape, were absent. This behaviour contrasted with the predictions of a statistical inference model with no priors, showing the strong role that shape-bias plays in human feature selection. It also contrasted with CNNs that (i) preferred to categorise objects based on non-shape features, and (ii) increased reliance on these non-shape features as they became more predictive. This was the case even when the CNN was pre-trained to have a shape-bias and the convolutional backbone was frozen. These results suggest that shape-bias has a different source in humans and CNNs: while learning in CNNs is driven by the statistical properties of the environment, humans are highly constrained by their previous biases, which suggests that cognitive constraints play a key role in how humans learn to recognise novel objects.

What do adversarial images tell us about human vision?

Deep convolutional neural networks (DCNNs) are frequently described as the best current models of human and primate vision. An obvious challenge to this claim is the existence of adversarial images that fool DCNNs but are uninterpretable to humans. However, recent research has suggested that there may be similarities in how humans and DCNNs interpret these seemingly nonsense images. We reanalysed data from a high-profile paper and conducted five experiments controlling for different ways in which these images can be generated and selected. We show human-DCNN agreement is much weaker and more variable than previously reported, and that the weak agreement is contingent on the choice of adversarial images and the design of the experiment. Indeed, we find there are well-known methods of generating images for which humans show no agreement with DCNNs. We conclude that adversarial images still pose a challenge to theorists using DCNNs as models of human vision.

The snow tire illusion: Different levels of perceptual assimilation across a single stimulus configuration.

While observing a specific traffic sign in the field, we noticed an apparent distortion of size and shape of the circle that contained the sign. This novel illusion manifests as a distortion of the horizontal compared to the vertical dimension of the sign. The illusion seems to be underlined by similar mechanisms to those in the Delboeuf illusion. The aim of our study was to determine the existence and magnitude of the snow tire illusion. We conducted two experiments using the method of constant stimuli. The first experiment was conducted on the standard sign, while in the second, the stimuli were rotated 90° counterclockwise. Both experiments consisted of three conditions: the snow tire, the ellipse, and the simple circle (control) conditions. The data showed a robust illusion effect for both the standard and rotated sign compared to the control condition, with a large majority of participants experiencing the illusion. The snow tire illusion seems to be a combination of assimilation mechanisms of different magnitudes. The assimilation is larger for one dimension of the sign, thus producing the shape distortion. The illusion may be a manifestation of a thus far undocumented non-uniform effect of assimilation on perceived size and shape.