Using machine learning to predict judgments on Western visual art along content-representational and formal-perceptual attributes.

Art research has long aimed to unravel the complex associations between specific attributes, such as color, complexity, and emotional expressiveness, and art judgments, including beauty, creativity, and liking. However, the fundamental distinction between attributes as inherent characteristics or features of the artwork and judgments as subjective evaluations remains an exciting topic. This paper reviews the literature of the last half century, to identify key attributes, and employs machine learning, specifically Gradient Boosted Decision Trees (GBDT), to predict 13 art judgments along 17 attributes. Ratings from 78 art novice participants were collected for 54 Western artworks. Our GBDT models successfully predicted 13 judgments significantly. Notably, judged creativity and disturbing/irritating judgments showed the highest predictability, with the models explaining 31% and 32% of the variance, respectively. The attributes emotional expressiveness, valence, symbolism, as well as complexity emerged as consistent and significant contributors to the models' performance. Content-representational attributes played a more prominent role than formal-perceptual attributes. Moreover, we found in some cases non-linear relationships between attributes and judgments with sudden inclines or declines around medium levels of the rating scales. By uncovering these underlying patterns and dynamics in art judgment behavior, our research provides valuable insights to advance the understanding of aesthetic experiences considering visual art, inform cultural practices, and inspire future research in the field of art appreciation.

In human-machine trust, humans rely on a simple averaging strategy.

With the growing role of artificial intelligence (AI) in our lives, attention is increasingly turning to the way that humans and AI work together. A key aspect of human-AI collaboration is how people integrate judgements or recommendations from machine agents, when they differ from their own judgements. We investigated trust in human-machine teaming using a perceptual judgement task based on the judge-advisor system. Participants ( n = 89 ) estimated a perceptual quantity, then received a recommendation from a machine agent. The participants then made a second response which combined their first estimate and the machine's recommendation. The degree to which participants shifted their second response in the direction of the recommendations provided a measure of their trust in the machine agent. We analysed the role of advice distance in people's willingness to change their judgements. When a recommendation falls a long way from their initial judgement, do people come to doubt their own judgement, trusting the recommendation more, or do they doubt the machine agent, trusting the recommendation less? We found that although some participants exhibited these behaviours, the most common response was neither of these tendencies, and a simple model based on averaging accounted best for participants' trust behaviour. We discuss implications for theories of trust, and human-machine teaming.

Large language models predict human sensory judgments across six modalities.

Determining the extent to which the perceptual world can be recovered from language is a longstanding problem in philosophy and cognitive science. We show that state-of-the-art large language models can unlock new insights into this problem by providing a lower bound on the amount of perceptual information that can be extracted from language. Specifically, we elicit pairwise similarity judgments from GPT models across six psychophysical datasets. We show that the judgments are significantly correlated with human data across all domains, recovering well-known representations like the color wheel and pitch spiral. Surprisingly, we find that a model (GPT-4) co-trained on vision and language does not necessarily lead to improvements specific to the visual modality, and provides highly correlated predictions with human data irrespective of whether direct visual input is provided or purely textual descriptors. To study the impact of specific languages, we also apply the models to a multilingual color-naming task. We find that GPT-4 replicates cross-linguistic variation in English and Russian illuminating the interaction of language and perception.

How people evaluate individuals who act morally prior to acting immorally: An examination of developmental change in moral evaluation, social preference, and prediction of moral behaviors.

Recent studies have enthusiastically examined the developmental origin of moral self-licensing, which is a tendency to act immorally after acting morally. However, it has not been considered enough how children evaluate personality traits of individuals who show moral licensing behavior and whether there is any developmental change in this evaluation. This study examined the developmental change in moral evaluation, social preference, and prediction of moral behaviors for moral licensing characters as well as moral or immoral characters. In total, 36 5- and 6-year-old children, 36 7- and 8-year-old children, and 58 university students participated in the study. The results revealed that 7- and 8-year-olds and adults evaluated moral licensing characters as more moral and likable than those who behave immorally, unlike 5- and 6-year-olds, who did not distinguish between the immoral and moral licensing characters. Importantly, 7- and 8-year-olds judged the moral licensing character as neutral in both moral evaluation and judgment of social preference, suggesting that they thought the immoral behavior was canceled out owing to prior moral behavior in the moral licensing character. However, adults still judged the moral licensing character as immoral and dislikable. Moreover, children's prediction of moral behavior for all characters showed the same tendency as moral evaluation, whereas adults' prediction was slightly different from their moral evaluation. Taken together, our findings revealed that the evaluation of individuals who show moral licensing behavior changed developmentally, and a moral licensing effect was found when evaluating others' moral traits from around 7 or 8 years of age.

Executive function as a mechanism for the emergence and expression of moral knowledge.

We outline two accounts by which executive function (EF) supports children's moral reasoning: an emergence and an expression account. The emergence account postulates that EF supports the development of moral concepts because it relates to how children navigate their early social environments and how well they can internalize moral messages. The expression account postulates that EF supports children's in-the-moment moral reasoning for complex moral situations. We present data from two studies with preschool children to assess each account. In support of the emergence account, EF longitudinally and positively predicted moral reasoning, but only for children who have experienced moderate forms of peer conflict. In support of the expression account, EF was only correlated with judgments that required the coordination of multiple pieces of information (i.e., retaliation and criterion judgments). We conclude that EF is an important cognitive mechanism of moral development and discuss various implications of these findings for both moral development and EF theory.

Both mOTS-words and pOTS-words prefer emoji stimuli over text stimuli during a lexical judgment task.

The visual word form area in the occipitotemporal sulcus (here OTS-words) is crucial for reading and shows a preference for text stimuli. We hypothesized that this text preference may be driven by lexical processing. Hence, we performed three fMRI experiments (n = 15), systematically varying participants' task and stimulus, and separately evaluated middle mOTS-words and posterior pOTS-words. Experiment 1 contrasted text with other visual stimuli to identify both OTS-words subregions. Experiment 2 utilized an fMRI adaptation paradigm, presenting compound words as texts or emojis. In experiment 3, participants performed a lexical or color judgment task on compound words in text or emoji format. In experiment 2, pOTS-words, but not mOTS-words, showed fMRI adaptation for compound words in both formats. In experiment 3, both subregions showed higher responses to compound words in emoji format. Moreover, mOTS-words showed higher responses during the lexical judgment task and a task-stimulus interaction. Multivariate analyses revealed that distributed responses in pOTS-words encode stimulus and distributed responses in mOTS-words encode stimulus and task. Together, our findings suggest that the function of the OTS-words subregions goes beyond the specific visual processing of text and that these regions are flexibly recruited whenever semantic meaning needs to be assigned to visual input.

Self-reference and emotional reaction drive aesthetic judgment.

Traditional philosophical inquiry, and more recently neuroscientific studies, have investigated the sources of artworks' aesthetic appeal. A substantial effort has been made to isolate the objective features contributing to aesthetic appreciation. While variables such as contrast or symmetry have been shown to robustly impact aesthetic judgment, they only account for a small portion of the intersubjective variability in aesthetic ratings. Recent multiprocess model of aesthetic appreciation could accommodate this finding by proposing that evaluative processes based on self-reference underpin the idiosyncrasy of aesthetic judgment. We tested this hypothesis in two behavioral studies, that were basically conceptual replications of our previous work, in which we took advantage of the self-reference effect on memory. We also tried to disentangle the role of self-reference and emotional reaction to artworks in guiding aesthetic judgments, by comparing an aesthetic judgment encoding condition to a self-reference condition (Study 1), and an emotional evaluation condition (Study 2). We show that artworks encoded in an aesthetic judgment condition exhibit a similar mnesic advantage compared to both the self-reference and the emotional evaluation encoding conditions. Moreover, retrospective emotional judgment correlates with both self-reference and aesthetic judgments ratings. These results suggest that a basic mechanism, appraisal of self-relevance, could ground aesthetic judgments.

How does musical rhythm influence grammatical processing at the neurophysiological level?

Numerous behavioral studies have demonstrated a rhythmic priming effect (RPE) on grammatical processing using grammaticality judgment tasks (GJT), where participants performed better following regular rhythmic sequences compared to baseline conditions or irregular rhythmic sequences (i.e. auditory rhythmic sequences with violated metrical structure). Only a few studies, however, have explored neurophysiological RPE in grammatical processing. Such neurophysiological investigations have been limited to GJT presented auditorily, have been primarily focused on the French- and German-speaking adult participants, and have rarely used baseline nonpriming conditions. The objective of the present study was to investigate neurophysiological correlates of the RPE in the GJT presented in visual modality. In the current study, we registered a 128-channel electroencephalogram while Russian-speaking adolescents performed a visual GJT, where each sentence was presented word by word in a self-paced manner. Before each experimental block, participants listened to regular rhythmic sequences, irregular rhythmic sequences, or silence. We observed that the late negativity in the event-related potential was larger for the ungrammatical condition compared to the grammatical condition only after the presentation of irregular rhythmic sequences. This effect, referred to as the N600 component in previous research, has been associated with increased cognitive complexity. In conclusion, results suggest that exposure to irregular rhythmic stimulation may lead to increased cognitive demand. This is attributed to the complexity associated with concurrently executing the GJT and managing rhythmic disruption, consequently increasing the strain on working memory resources.

Design and evaluation of crowdsourcing platforms based on users' confidence judgments.

Crowdsourcing deals with solving problems by assigning them to a large number of non-experts called crowd using their spare time. In these systems, the final answer to the question is determined by summing up the votes obtained from the community. The popularity of these systems has increased by facilitating access for community members through mobile phones and the Internet. One of the issues raised in crowdsourcing is how to choose people and how to collect answers. Usually, users are separated based on their performance in a pre-test. Designing the pre-test for performance calculation is challenging; The pre-test questions should be selected to assess characteristics in individuals that are relevant to the main questions. One of the ways to increase the accuracy of crowdsourcing systems is by considering individuals' cognitive characteristics and decision-making models to form a crowd and improve the estimation of their answer accuracy to questions. People can estimate the correctness of their responses while making a decision. The accuracy of this estimate is determined by a quantity called metacognition ability. Metacoginition is referred to the case where the confidence level is considered along with the answer to increase the accuracy of the solution. In this paper, by both mathematical and experimental analysis, we would answer the following question: Is it possible to improve the performance of a crowdsourcing system by understanding individuals' metacognition and recording and utilizing users' confidence in their answers?

Do claims about certainty make estimates less certain?

Predictions and estimates are sometimes qualified as certain. This epistemic marker occupies a privileged position at the top of scales of verbal probability expressions, reflecting probabilities close to 1. But such statements have rarely been compared to plain, unqualified statements in which certainty is not mentioned. We examined in nine studies (N = 2784) whether statements explicitly claimed to be certain are perceived as (1) more (or less) credible, (2) more (or less) precise, and (3) more (or less) strongly based upon evidence, compared to plain, unmarked declarative statements. We find, in apparent contrast with assumptions made by the standard scales, that "certain" are often judged to be less trustworthy, less reliable, and held with lower confidence than unmarked statements. Plain, declarative statements are further assumed to be more precise, while certainty implies that more extreme outcomes are possible. When it is certain that Henry made four errors, it is clear he did not commit less than four, but he might have committed five errors or more. Thus certainty can indicate lower bounds of an interval whose upper bounds are not defined, and certainty statements are consequently more ambiguous than estimates that do not mention certainty. At least-interpretations of certainty affect the interpretation of options in risky choice problems, where "200 lives will be saved" was deemed by a majority to mean exactly 200, while "it is certain that 200 will be saved", could mean 200-600 lives. We also find that credibility is affected by type of certainty, with impersonal certainty ("it is certain") perceived to be more accurate and persuasive than personal certainty ("I am certain"), especially in predictions of future events. Moreover, mentions of certainty can reveal that that a speaker's estimate is based on subjective judgments and guesswork rather than upon objective evidence. These findings have implications for communication. Estimates can appear more consensual when claims of certainty are omitted. To convey certainty it may be better not to mention that one is certain.

Moral preference reversals: Violations of procedure invariance in moral judgments of sacrificial dilemmas.

In this research, we examine whether moral judgments sometimes violate the normative principle of procedure invariance - that is, whether normatively equivalent elicitation tasks can result in different judgment patterns. Specifically, we show that the relative morality of two actions can reverse across evaluation modes and elicitation tasks, mirroring preference reversals in consumer behavior. Across six studies (five preregistered, total N = 719), we provide evidence of three reversals of moral judgments of sacrificial dilemmas. First, directly killing one person to save many others was rated as morally worse than indirectly killing one person via an intervening mechanism in order to save a few others in separate evaluation, but this difference reversed in joint evaluation, in both between-subjects (Studies 1a and 1b) and within-subjects (Study 2) designs. Next, directly killing one person to save many others was judged as morally better than indirectly killing one person to save a few others more often in matching than in choice (Study 3) and rating (Study 4), between-subjects. Lastly, we replicate the results of Studies 3 and 4 within-subjects and show that susceptibility to these moral preference reversals is correlated with Faith in Intuition (Study 5). The present research introduces a new methodological approach to moral psychology, demonstrates that moral judgments can fully reverse across tasks, and supports an emerging view that moral judgments, like consumer preferences, are at least sometimes constructed in the moment, relative to the context and task at hand.

Metacognition bridges experiences and beliefs in sense of agency.

Cognitive scientists differentiate the "minimal self" - subjective experiences of agency and ownership in our sensorimotor interactions with the world - from declarative beliefs about the self that are sustained over time. However, it remains an open question how individual sensory experiences of agency are integrated into the belief ofbeing an agent.We administered a sensorimotor task to measure subjects' (n = 195) propensity to classify stimuli as self-caused and metacognitive monitoring of such judgements, and we compared these behavioral metrics to declarative beliefs about their agency. Subjects who were less sensitive to control cues also reported more negative agency beliefs, though positive beliefs were not clearly correlated with any sensorimotor measure. Importantly, this relationship between first-order sensitivity and declarative beliefs essentially disappears when controlling for metacognitive sensitivity. Results suggest agency beliefs are not related directly to the propensity to make positive agency judgements but are connected through introspective access.

Bird song comparison using deep learning trained from avian perceptual judgments.

Our understanding of bird song, a model system for animal communication and the neurobiology of learning, depends critically on making reliable, validated comparisons between the complex multidimensional syllables that are used in songs. However, most assessments of song similarity are based on human inspection of spectrograms, or computational methods developed from human intuitions. Using a novel automated operant conditioning system, we collected a large corpus of zebra finches' (Taeniopygia guttata) decisions about song syllable similarity. We use this dataset to compare and externally validate similarity algorithms in widely-used publicly available software (Raven, Sound Analysis Pro, Luscinia). Although these methods all perform better than chance, they do not closely emulate the avian assessments. We then introduce a novel deep learning method that can produce perceptual similarity judgements trained on such avian decisions. We find that this new method outperforms the established methods in accuracy and more closely approaches the avian assessments. Inconsistent (hence ambiguous) decisions are a common occurrence in animal behavioural data; we show that a modification of the deep learning training that accommodates these leads to the strongest performance. We argue this approach is the best way to validate methods to compare song similarity, that our dataset can be used to validate novel methods, and that the general approach can easily be extended to other species.

Unveiling the power of optimism: Exploring behavioral and neuromolecular correlates of alcohol seeking and drinking in rats with biased judgement.

Alcohol use disorder (AUD) is a common psychiatric condition with substantial global mortality. Despite extensive research into its pathophysiology, the cognitive predispositions driving alcohol dependence are less understood. This study explores whether biased cognition, specifically traits of optimism and pessimism, predicts susceptibility to alcohol-seeking behaviors using an animal model. Rats were initially tested for judgement bias through Ambiguous Cue Interpretation tests. Those identified as 'optimistic' or 'pessimistic' were further examined for their tendency to escalate alcohol intake using the intermittent access 2-bottle choice (2BC) paradigm. Additionally, we assessed how judgement bias influenced the development of compulsive alcohol-seeking behavior in a Seeking-Taking (ST) and Seeking-Taking Punishment tasks, alcohol-seeking motivation in the Progressive Ratio Schedule of Reinforcement paradigm, the speed of extinction, and reinstatement after abstinence. Neurochemical analyses were conducted to investigate trait-specific differences in neurotransmitter-related gene expression and receptor densities in the brain. We used TaqMan Gene Expression Array Cards to analyze expression levels of genes linked to serotonergic, dopaminergic, glutamatergic, and GABAergic pathways, and alcohol metabolism in various brain regions. Receptor densities for 5-HT1A, 5-HT2A, and D2 were measured using autoradiography analysis. Behaviorally, 'optimistic' rats showed significantly lower alcohol consumption in the 2BC paradigm compared to 'pessimistic' rats. This lowered intake correlated with decreased monoamine oxidase-A (Maoa) expression in the medial prefrontal cortex (mPFC) and increased metabotropic glutamate receptor 2 (Grm2) expression in the amygdala (Amy). Additionally, we observed significant interactions between judgement bias and alcohol intake in the expression of several genes in the mPFC, nucleus accumbens (Nacc), orbitofrontal cortex (OFC), and Amy, as well as in 5-HT2A receptor binding in the Nacc. Overall, these results suggest that optimism is linked to lower alcohol consumption and related neurochemical changes, indicating a potential cognitive mechanism in AUD risk.

Visual perspective taking and action understanding.

Understanding what others are doing is a fundamental aspect of social cognition and a skill that is arguably linked to visuospatial perspective taking (VPT), the ability to apprehend the spatial layout of a scene from another's perspective. Yet, with few and notable exceptions, action understanding and VPT are rarely studied together. Participants (43 females, 37 males) made judgements about the spatial layout of objects in a scene from the perspective of an avatar who was positioned at 0°, 90°, 270° or 180° relative to the participant. In a variant of a traditional VPT task, the avatar either interacted with the objects in the scene, by pointing to or reaching for them, or was present but did not engage with the objects. Although the task was identical across all conditions - to say whether a target object is to the right or left of a control object - we show that the avatar's actions modulates performance. Specifically, participants were more accurate when the avatar engaged with the target object, and correspondingly, less accurate and slower when the avatar interacted with the control objects. As these effects were independent of the angular disparity between participant and avatar perspectives, we conclude that action understanding and VPT are likely linked via the rapid deployment of two separate cognitive mechanisms. All participants provided a measure of self-reported empathy and we show that response times decrease with increasing empathy scores for female but not for male participants. However, within the range of 'typical' empathy scores, defined here as the interquartile range where 50 % of the data lie, females were faster than males. These findings lend further insight into the relationship between spatial and social perspective taking.

The good, the rich, and the powerful: How young children compensate victims of moral transgressions depending on moral character, wealth, and social dominance.

Theories of justice suggest that it serves two main purposes: punishment and restoration. Although punishment emerges early and has been well-documented, little is known about the contexts in which young children engage in restorative practices like compensation for victims. The current study investigated whether children's engagement in compensation and punishment (which often involve a redistribution of resources) was sensitive to characteristics of the perpetrator and victim known to shape distributive justice decisions (decisions about how resources should be distributed), such as social dominance, resource inequality, and moral character. A total of 54 children aged 3 to 7 years completed a series of moral judgment experiments. Each experiment featured interactions between a perpetrator and a victim, ending with the perpetrator stealing the victim's toy. In Experiment 1 (N = 44), social dominance did not affect punishment or compensation overall, but older children compensated the dominant victim (but not the subordinate victim) less than younger children. In Experiment 2 (N = 42), children compensated the poor victim more than the rich victim, but they did not punish the rich perpetrator more than the poor perpetrator. In Experiment 3 (N = 45), children compensated the victim with a good moral character more than the victim with a bad moral character, and the victim's moral character did not influence punishment. Altogether, these findings offer new insights into how children resort to compensation for victims as a complement to, rather than an alternative to, punishment.

Ethical choice reversals.

Understanding the systematic ways that human decision making departs from normative principles has been important in the development of cognitive theory across multiple decision domains. We focus here on whether such seemingly "irrational" decisions occur in ethical decisions that impose difficult tradeoffs between the welfare and interests of different individuals or groups. Across three sets of experiments and in multiple decision scenarios, we provide clear evidence that contextual choice reversals arise in multiples types of ethical choice settings, in just the way that they do in other domains ranging from economic gambles to perceptual judgments (Trueblood et al., 2013; Wedell, 1991). Specifically, we find within-participant evidence for attraction effects in which choices between two options systematically vary as a function of features of a third dominated and unchosen option-a prima facie violation of rational choice axioms that demand consistency. Unlike economic gambles and most domains in which such effects have been studied, many of our ethical scenarios involve features that are not presented numerically, and features for which there is no clear majority-endorsed ranking. We provide empirical evidence and a novel modeling analysis based on individual differences of feature rankings within attributes to show that such individual variations partly explains observed variation in the attraction effects. We conclude by discussing how recent computational analyses of attraction effects may provide a basis for understanding how the observed patterns of choices reflect boundedly rational decision processes.

Goal inference in moral narratives.

We often form beliefs about others based on narratives they tell about their own moral actions. When constructing such moral narratives, narrators balance multiple goals, such as conveying accurate information about what happened ('informational goals') and swaying audiences' impressions about their moral characters ('reputational goals'). Here, we ask to what extent audiences' detection of narrators' reputational goals guide or prevent them from making moral character judgments intended by narrators. Across two pre-registered experiments, audiences read narratives written by real narrators about their own moral actions. Each narrator was incentivized to write about the same action twice while trying to appear like a morally good or bad person (positive and negative reputational goals). Audiences detected narrators' reputational goals with high accuracy and made judgments about moral character that aligned with narrators' goals. However, audiences were more suspicious toward positive than negative reputational goals, requiring more evidence of high informational goals. These results demonstrate how audiences' inferences of reputational goals can both support and hinder narrators: accurate goal recognition increases the chance that audiences will make judgments intended by narrators, but inferred positive reputational goals can lead to doubts about accuracy. More generally, this provides a novel approach to studying how moral information about people is transmitted through naturalistic narratives.