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. (shrink)

Decades of psychological research have been aimed at modeling how people learn features and categories. The empirical validation of these theories is often based on artificial stimuli with simple representations. Recently, deep neural networks have reached or surpassed human accuracy on tasks such as identifying objects in natural images. These networks learn representations of real‐world stimuli that can potentially be leveraged to capture psychological representations. We find that state‐of‐the‐art object classification networks provide surprisingly accurate predictions of human similarity judgments for (...) natural images, but they fail to capture some of the structure represented by people. We show that a simple transformation that corrects these discrepancies can be obtained through convex optimization. We use the resulting representations to predict the difficulty of learning novel categories of natural images. Our results extend the scope of psychological experiments and computational modeling by enabling tractable use of large natural stimulus sets. (shrink)

More than 50 years ago, Bongard introduced 100 visual concept learning problems as a challenge for artificial vision systems. These problems are now known as Bongard problems. Although they are well known in cognitive science and artificial intelligence, only very little progress has been made toward building systems that can solve a substantial subset of them. In the system presented here, visual features are extracted through image processing and then translated into a symbolic visual vocabulary. We introduce a formal language (...) that allows representing compositional visual concepts based on this vocabulary. Using this language and Bayesian inference, concepts can be induced from the examples that are provided in each problem. We find a reasonable agreement between the concepts with high posterior probability and the solutions formulated by Bongard himself for a subset of 35 problems. While this approach is far from solving Bongard problems like humans, it does considerably better than previous approaches. We discuss the issues we encountered while developing this system and their continuing relevance for understanding visual cognition. For instance, contrary to other concept learning problems, the examples are not random in Bongard problems; instead they are carefully chosen to ensure that the concept can be induced, and we found it helpful to take the resulting pragmatic constraints into account. (shrink)

When a circular coin is rotated in depth, is there any sense in which it comes to resemble an ellipse? While this question is at the center of a rich and divided philosophical tradition (with some scholars answering affirmatively and some negatively), Morales et al. (2020, 2021) took an empirical approach, reporting 10 experiments whose results favor such perspectival similarity. Recently, Burge and Burge (2022) offered a vigorous critique of this work, objecting to its approach and conclusions on both philosophical (...) and empirical grounds. Here, we answer these objections on both fronts. We show that Burge and Burge’s critique rests on misunderstandings of Morales et al.’s claims; of the relation between the data and conclusions; and of the philosophical context in which the work appears. Specifically, Burge and Burge attribute to us a much stronger (and stranger) view than we hold, involving the introduction of “a new entity” located “in some intermediate position(s) between the distal shape and the retinal image.” We do not hold this view. Indeed, once properly understood, most of Burge and Burge’s objections favor Morales et al.’s claims rather than oppose them. Finally, we discuss several questions that remain unanswered, and reflect on a productive path forward on these issues of foundational scientific and philosophical interest. (shrink)

Although discriminative deep neural networks are currently dominant in cognitive modeling, we suggest that capacity-limited, generative models are a promising avenue for future work. Generative models tend to learn both local and global features of stimuli and, when properly constrained, can learn componential representations and response biases found in people's behaviors.