Do Humans and Deep Convolutional Neural Networks Use Visual Information Similarly for the Categorization of Natural Scenes?

The investigation of visual categorization has recently been aided by the introduction of deep convolutional neural networks (CNNs), which achieve unprecedented accuracy in picture classification after extensive training. Even if the architecture of CNNs is inspired by the organization of the visual brain, the similarity between CNN and human visual processing remains unclear. Here, we investigated this issue by engaging humans and CNNs in a two-class visual categorization task. To this end, pictures containing animals or vehicles were modified to contain only low/high spatial frequency (HSF) information, or were scrambled in the phase of the spatial frequency spectrum. For all types of degradation, accuracy increased as degradation was reduced for both humans and CNNs; however, the thresholds for accurate categorization varied between humans and CNNs. More remarkable differences were observed for HSF information compared to the other two types of degradation, both in terms of overall accuracy and image-level agreement between humans and CNNs. The difficulty with which the CNNs were shown to categorize high-passed natural scenes was reduced by picture whitening, a procedure which is inspired by how visual systems process natural images. The results are discussed concerning the adaptation to regularities in the visual environment (scene statistics); if the visual characteristics of the environment are not learned by CNNs, their visual categorization may depend only on a subset of the visual information on which humans rely, for example, on low spatial frequency information.     

Interactions entre règles sociales de santé publique : étude des liens entre le port du masque et les règles routières en temps de crise sanitaire

IntroductionThe health crisis that France has experienced since the beginning of 2020 has justified the introduction of new health rules requiring changes in behaviour. The context of the implementation of these health rules, instituted very recently and for a limited period of time, raises the question of their perception and respect among the population.ObjectiveMobilising the social developmental psychology, the objective of this article is to explore inter-individual differences in the perception of these health rules measured through the wearing of masks. It aims also at studying the links between these perceptions with the experience of the health context and the perception of other legal public health rules, apprehended through some road safety rules.MethodThe questionnaire survey, carried out during the second confinement with a representative sample of 4999 people, made it possible to measure the experience of the health context, the internalisation of the rules and their justifications, the usefulness of the rules and the perception of the legal system.ResultsThe results show the links between internalisation, the type of arguments used to justify it for wearing a mask and for road rules respectively, as well as the links between internalisation and the perception of the legal system. They also show that women use moral and prudential arguments more to justify their internalisation of rules and that they have internalised the traffic rules (but not the wearing of masks) more than men.ConclusionThese results show that the perception of the rule of wearing a mask is more related to the general perception that individuals have of the legal system and other legal public health rules involving the relationship with others than to the context in which these rules were put in place.     

Gender differences in implicit gender self‐categorization lead to stronger gender self‐stereotyping by women than by men

It was hypothesized that, in natural group contexts, low‐status in‐group membership would be highly accessible, whereas membership to high‐status groups would not. Therefore, gender group membership was predicted to be more accessible for women than for men. It was further hypothesized that the high accessibility of gender group membership would lead to stronger self‐stereotyping for women than for men. To measure the accessibility of gender group membership, participants performed a Gender Self‐Categorization Implicit Association Test (Studies 1 and 2), measuring the strength of automatic associations between the self and the gender in‐group. Participants also performed a Self‐Stereotyping Implicit Association Test (Study 2), assessing the strength of automatic associations between the self and the stereotypical traits of the in‐group. As expected, implicit gender self‐categorization and implicit gender self‐stereotyping were stronger for women than for men. Importantly, implicit gender self‐categorization mediated the relation between gender and self‐stereotyping. Therefore, implicit gender self‐categorization was the mechanism underlying stronger implicit self‐stereotyping by women. Copyright © 2012 John Wiley & Sons, Ltd.     

The question shapes the answer: the neural correlates of task differences reveal dynamic semantic processing

Task effects in semantic processing were investigated by contrasting the neural activation associated with two semantic categorization tasks (SCT) using event-related fMRI. The two SCTs involved different decision categories: is it an animal? vs. is it a concrete thing? Participants completed both tasks and, across participants, the same core set of items were presented in both tasks. Results showed task differences in the neural activation associated with these items: in the animal SCT there was greater activation in a number of frontal and temporal regions, including left superior and middle temporal gyri, while in the concrete SCT there was greater activation in left medial frontal gyrus and bilaterally in the precentral gyri. These results are interpreted as evidence of top-down modulation of semantic processing; participants make adjustments to optimize performance in a given task and these adjustments have consequences for the activation observed.     

Familiarity increases the accuracy of categorizing male sexual orientation

Although individual differences are known to influence numerous aspects of social perception, such as person memory and individuation, little is known about how such variations may affect social categorization. Extending prior research, the present study tested one potential moderator: familiarity with group members. Specifically, straight participants (n = 84) reported their real-life experiences with gay men and categorized faces as gay or straight. Results showed that participants who reported greater familiarity with gay men were significantly more sensitive, or accurate, in judging the sexual orientations of men from their faces. These results are discussed in terms of their theoretical implications for social perception and future research directions are outlined.     

Emotional,behavioral, and cognitive factors that differentiate obsessive-compulsive disorder and other anxiety disorders in youth

Abstract

The current study examined specific emotional, behavioral, and cognitive variables that may distinguish obsessive-compulsive disorder (OCD) from generalized anxiety disorder (GAD), social phobia (SoP), and separation anxiety disorder (SAD) in youth. Youth with OCD (n=26) and other anxiety disorders (ADs; n=31), aged 7–12 years (56.1% males), and their parents participated. The study compared the two anxious groups on levels of emotional, behavioral, and cognitive functioning, as well as impairment associated with the disorder. Results indicated that in comparison to youth with GAD, SoP, or SAD, youth with OCD were found to have poorer emotion regulation skills, as well as greater oppositionality, cognitive problems/inattention, and parent impairment associated with the disorder. The findings suggest that there are unique characteristics of OCD that may differentiate this disorder from other ADs in youth. Potential clinical implications and directions for future research are discussed.     

When do ratings implicate perception versus judgment? The “overgeneralization test” for top-down effects

ABSTRACT

When looking at an object—say, a banana—we can both directly perceive its visual qualities (e.g., its size) and also make higher-level judgments about its visual and non-visual properties (e.g., not only its size, but also its cost). Suppose you obtain a rating of a property such as size. Does that rating implicate seeing or merely higher-level judgment? The answer often matters a great deal -- e.g., determining whether such ratings imply “top-down” effects of cognition on perception. Too often, however, this distinction is ignored in empirical investigations of such effects. Here we suggest a simple test for when such ratings can be used to implicate perception: whenever the very same experiment “overgeneralizes” to an unambiguously non-perceptual factor, the results cannot be used to draw implications about perception, per se. As a case study, we investigate an empirical report alleging that conservatives perceive Barack Obama as darker skinned than liberals do. Two simple experiments show that the very same effects, measured via the same “representativeness” ratings, obtain with unambiguously non-perceptual and even silly factors (involving bright red horns vs. halos, rather than brightness differences). We suggest that this renders such methods unable to implicate visual processing and we recommend that “overgeneralization tests” of this type always be conducted in such contexts.     

Infant directed speech and the development of speech perception: Enhancing development or an unintended consequence?

Infant directed speech (IDS) is a speech register characterized by simpler sentences, a slower rate, and more variable prosody. Recent work has implicated it in more subtle aspects of language development. Kuhl et al. (1997) demonstrated that segmental cues for vowels are affected by IDS in a way that may enhance development: the average locations of the extreme “point” vowels (/a/, /i/ and /u/) are further apart in acoustic space. If infants learn speech categories, in part, from the statistical distributions of such cues, these changes may specifically enhance speech category learning. We revisited this by asking (1) if these findings extend to a new cue (Voice Onset Time, a cue for voicing); (2) whether they extend to the interior vowels which are much harder to learn and/or discriminate; and (3) whether these changes may be an unintended phonetic consequence of factors like speaking rate or prosodic changes associated with IDS. Eighteen caregivers were recorded reading a picture book including minimal pairs for voicing (e.g., beach/peach) and a variety of vowels to either an adult or their infant. Acoustic measurements suggested that VOT was different in IDS, but not in a way that necessarily supports better development, and that these changes are almost entirely due to slower rate of speech of IDS. Measurements of the vowel suggested that in addition to changes in the mean, there was also an increase in variance, and statistical modeling suggests that this may counteract the benefit of any expansion of the vowel space. As a whole this suggests that changes in segmental cues associated with IDS may be an unintended by-product of the slower rate of speech and different prosodic structure, and do not necessarily derive from a motivation to enhance development.     

Measuring category intuitiveness in unconstrained categorization tasks

What makes a category seem natural or intuitive? In this paper, an unsupervised categorization task was employed to examine observer agreement concerning the categorization of nine different stimulus sets. The stimulus sets were designed to capture different intuitions about classification structure. The main empirical index of category intuitiveness was the frequency of the preferred classification, for different stimulus sets. With 169 participants, and a within participants design, with some stimulus sets the most frequent classification was produced over 50 times and with others not more than two or three times. The main empirical finding was that cluster tightness was more important in determining category intuitiveness, than cluster separation. The results were considered in relation to the following models of unsupervised categorization: DIVA, the rational model, the simplicity model, SUSTAIN, an Unsupervised version of the Generalized Context Model (UGCM), and a simple geometric model based on similarity. DIVA, the geometric approach, SUSTAIN, and the UGCM provided good, though not perfect, fits. Overall, the present work highlights several theoretical and practical issues regarding unsupervised categorization and reveals weaknesses in some of the corresponding formal models.     

The role of problem solving in complex intraverbal repertoires

We examined whether typically developing preschoolers could learn to use a problem-solving strategy that involved self-prompting with intraverbal chains to provide multiple responses to intraverbal categorization questions. Teaching the children to use the problem-solving strategy did not produce significant increases in target responses until problem solving was modeled and prompted. Following the model and prompts, all participants showed immediate significant increases in intraverbal categorization, and all prompts were quickly eliminated. Use of audible self-prompts was evident initially for all participants, but declined over time for 3 of the 4 children. Within-session response patterns remained consistent with use of the problem-solving strategy even when self-prompts were not audible. These findings suggest that teaching and prompting a problem-solving strategy can be an effective way to produce intraverbal categorization responses.