Anxiety is associated with altered patterns of attention to images and objects, but how it influences the perception of continuous experiences remains underexplored. Here, we examined how anxiety influences the segmentation of continuous experience into discrete events. Using a large-scale open-access dataset and data-driven methods to detect neural event boundaries, we compared neural event segmentation between healthy adults with high (N=60) and low trait-level anxiety (N=60) as they viewed a short suspenseful film. While the overall temporal event segmentation hierarchy was preserved across groups, higher-anxiety individuals had more idiosyncratic boundaries, exhibiting reduced alignment to both low-anxiety individuals and one another. This divergence was the most pronounced in the dorsal attention and frontoparietal control networks, suggesting that anxiety-related disruptions in top-down attentional control may contribute to less consistent segmentation. Furthermore, boundary variability in anxious participants was higher during moments rated as more anxiety-provoking by a large language model, controlling for arousal, valence, and low-level visual features. Together, these findings suggest that idiosyncratic event models in anxious individuals result in non-normative organization of continuous experience, leading to more individualized and inconsistent event boundaries. Such variability in segmentation may have downstream consequences for prediction, memory, and social cognition, offering a potential neural mechanism by which anxiety influences the interpretation and recall of everyday experiences. Our findings highlight trait-level variability as a lens through which the brain parses information and offers a neurocomputational framework for investigating individual differences in naturalistic event cognition. ### Competing Interest Statement The authors have declared no competing interest.

Resting-state functional connectivity (rsFC)-brain connectivity observed when people rest with no external tasks-predicts individual differences in behaviour. Yet, rest is not idle; it involves streams of thoughts. Are these ongoing thoughts reflected in FC and do they contribute to the relationship between rsFC and behaviour? Here, to test this question, we used an annotated rest paradigm where participants rated and verbally described their thoughts after each rest period during functional MRI (N = 60). Our findings revealed rich and idiosyncratic thoughts across individuals. Similarity in thoughts was associated with more similar FC patterns within and across individuals. In addition, both thought ratings and topics could be decoded from FC. Furthermore, neuromarkers of these thoughts generalized to unseen individuals in the Human Connectome Project dataset (N = 908), where decoded thought patterns during rest predicted positive vs negative trait-level individual differences. Together, our findings reveal that ongoing thoughts at rest are reflected in brain dynamics and these network patterns predict everyday cognition and experiences. Understanding subjective in-scanner experiences is thus crucial in characterizing the relationship between individual differences in functional brain organization and behaviour.

Teachers' hand movements during instruction can influence how children learn mathematics, but not all movements are equally effective. Gesture-based instruction on problems such as 4 + 2 + 5 = __ + 5-where the teacher places a V-shaped hand under the 4 and 2 and then points to the blank ("grouping strategy")-promotes learning better than action-based instruction-where the teacher manipulates magnetic numbers to mimic the gestures. How do hand movements facilitate learning? We tested 8- to 10-year-old children (n = 73) using functional near-infrared spectroscopy to measure neural activity as they watched gesture-based or action-based videotaped lessons. Gesture-based instruction elicited greater intersubject neural synchrony in motor cortex and angular gyrus, a region implicated in arithmetic processing. Critically, synchrony in the right angular gyrus during gesture instruction predicted learning gains, whereas synchrony during action-based instruction did not. Our work demonstrates how gestures foster shared representations in brain regions supporting arithmetic reasoning, offering a neural basis for gesture's behavioral advantages in mathematics education. These findings highlight functional near-infrared spectroscopy as a powerful tool for capturing how mathematical learning unfolds in children.

We make sense of everyday events by reasoning about their underlying causes. When we connect causal links between events separated in time, we often experience a sudden feeling of "aha!", or insight. To understand its cognitive and neural mechanisms, we designed an fMRI study in which participants watched a temporally scrambled TV episode. Participants pressed an "aha" button whenever they understood something new and explained their reasoning. More than 40% of insight explanations reference past events causally related to the current event. Neural patterns representing these causally related past events are reinstated across the cortex. This neural reinstatement drives sudden shifts in cortical activity patterns ~2 s prior to aha button presses, reflecting updates in situational representation. Moreover, distributed brain areas represent causally related events with similar neural patterns, beyond shared semantic or perceptual features. Together, the study suggests that we comprehend events by retrieving causally related past events, followed by updating neural patterns at moments of insight.

Abstract Arousal is often invoked to explain state-dependent variability in attention, emotion, memory, and decision-making. However, the term is used inconsistently, referring variously to affective experience, autonomic activation, and states of wakefulness. The extent to which different operationalizations of arousal reflect a shared neurobiological basis remains unclear, limiting efforts to unify findings across studies. We applied dynamic connectome predictive models to five fMRI datasets spanning naturalistic movie watching, story listening, wakeful rest, and sleep. We derived measures of affective, autonomic, and wakefulness arousal from subjective ratings, pupil dilation, and EEG, respectively. Models trained to predict arousal in one dataset generalized across datasets, indicating that dynamic connectivity captures shared arousal-related dynamics across measures and task contexts. The models also predicted manually scored sleep stages, suggesting that they capture arousal dynamics that extend to the graded fluctuations in arousal during sleep. Decoded arousal dynamics during movie-viewing predicted how well participants later recalled movie events, reproducing classic arousal-dependent memory enhancement effects. Model predictions were supported by overlapping functional connections, including a subset shared across all models. The largest proportion of shared connections was between the salience and somatomotor networks, suggesting that increased coordination between salience detection and action readiness may be a common feature across multiple operationalizations of arousal. Together, these results are consistent with a connectome-based neural reference space for arousal, in which different varieties share a core set of predictive connections. Our findings offer a quantitative framework for integrating arousal-related findings across tasks and modalities.

Affective polarization, defined as the dislike between opposing political groups, is a growing global threat. While much of the focus has been on partisan identities, political divisions may also be driven by affective divergence around political issues, where partisans express opposing feelings toward topics they disagree about. To compare identity-based and issue-based affective alignment, we used word embeddings to analyze two large datasets comprising ~300 million comments from partisan Reddit communities and ~7 million articles from partisan news outlets. We first quantified affective alignment by measuring the valence associations of identity and issue words. In both datasets, affective alignment was greater around political issues than around partisan identities. To validate these findings using a context-sensitive approach, we also used a large language model to rate the valence of identity and issue words in Reddit comments. We again observed stronger affective agreement around issues than identities. These results reveal that even though partisans hold strong negative attitudes toward opposing partisans, the emotional divide around political issues is less pronounced, suggesting opportunities for bridging partisan differences through issue-focused dialog. Our study offers scalable, quantitative tools for understanding the emotional dimensions of political polarization and highlighting pathways to reduce its impact.

Intermittent explosive disorder (IED) is associated with impulsive aggression in ambiguous social contexts. Prior neuroimaging studies have treated IED as a homogenous group, but identical social situations may elicit divergent responses across IED individuals. Here, we test the hypothesis that IED is characterized by idiosyncratic neural responses to social cues during naturalistic social-emotional processing. IED individuals and healthy controls completed a validated paradigm where they were presented with video vignettes of interpersonal interactions while undergoing fMRI. We computed the intersubject correlation (ISC) in neural time courses between pairs of participants to quantify neural similarity, and assessed whether similarity differed between Healthy-Healthy and IED-IED dyads using Bayesian multilevel models, controlling for self-reported emotional responses and intention attributions for each vignette. Healthy-Healthy dyads showed significantly higher ISC than IED-IED dyads, indicating that neural responses to the videos were similar among healthy participants, but idiosyncratic in IED individuals. These effects were observed in regions in the default mode and salience networks, including the precuneus, medial prefrontal cortex, superior temporal sulcus, insula, and dorsal anterior cingulate cortex. Individuals with IED exhibited idiosyncratic neural responses during naturalistic social-emotional processing, even after accounting for differences in emotional reaction and intention attribution. This neural idiosyncrasy may reflect atypical integration of social cues, giving rise to maladaptive interpretations and impulsive aggression. Assessing neural synchrony during ecologically valid paradigms offers a promising tool for identifying neural markers of interpersonal dysfunction and informing targeted interventions.

Abstract Updating impressions of others is essential to navigating social life. As we get to know an individual, we update our impressions of them in accordance with new information. How the brain dynamically revises impressions in real time under naturalistic conditions remains unclear. Here, we address this question using functional magnetic resonance imaging (fMRI) and natural language analysis in a naturalistic social cognition paradigm. Across 10 runs, participants viewed a character-driven TV episode, reported moments of insight, and described their impressions of the characters. Results reveal that impressions progressively evolved over time. Individuals with more similar existing impressions exhibit greater neural synchrony during movie-watching, which predicts convergence in post-movie impressions. Neural synchrony in the right superior temporal sulcus (STS) mediates the influence of initial similarity on later alignment. Insight moments accompany neural pattern shifts widespread across cortical regions, including the temporoparietal junction (TPJ), dorsomedial prefrontal cortex (dmPFC) and STS, and the magnitude of the shifts tracks the degree of impression updating. Specifically, distinct forms of insight selectively update complementary components of impressions: character insight updates person-centered representations, whereas non-character insight shapes social-event structure. Together, these findings show that people update their impressions of others through dynamic shifts in distributed brain activity patterns at moments of insight, providing a novel and ecologically grounded neural account of social cognition. Significance statement People do not simply form first impressions and keep them. As we learn new things about others, our judgments change—sometimes gradually, sometimes in a flash of insight. By tracking brain activity while people watched a TV show and described how their impressions of the characters evolved, we found that people who saw a character in more similar ways also show more similar brain responses when watching them onscreen, which in turn leads to convergence in subsequent impressions. Such updating is closely tied to “aha” moments, when a person is suddenly understood in a new way. These moments were marked by rapid shifts in the brain representational patterns, and larger shifts predicted greater changes in impressions. These findings offer a more naturalistic account of how the brain revises our understanding of other people in everyday life.

Hostile attribution bias is the tendency to interpret social interactions as intentionally hostile. The Ambiguous Intentions Hostility Questionnaire (AIHQ) is a commonly used instrument to measure hostile attribution bias and includes open-ended questions where participants describe the perceived intentions behind a negative social situation and how they would respond. While these questions provide insights into the contents of hostile attributions, they require time-intensive scoring by human raters. In this study, we assessed whether large language models can automate the scoring of AIHQ open-ended responses. We used a previously collected data set in which individuals with traumatic brain injury (TBI) and non-TBI controls completed the AIHQ and had their open-ended responses rated by trained human raters. We used half of these responses to fine-tune the two models on human-generated ratings and tested the fine-tuned models on the remaining half of AIHQ responses. Results showed that model-generated ratings aligned with human ratings for both attributions of hostility and aggression responses, with fine-tuned models showing higher alignment. This alignment was consistent across ambiguous, intentional, and accidental scenario types and replicated previous findings on group differences in attributions of hostility and aggression responses between TBI and non-TBI groups. The fine-tuned models also generalized well to an independent nonclinical data set. To support broader adoption, we provide an accessible scoring interface that includes both local and cloud-based options. Together, our findings suggest that large language models can streamline AIHQ scoring in both research and clinical contexts, revealing their potential to facilitate psychological assessments across different populations. (PsycInfo Database Record (c) 2026 APA, all rights reserved).

Most people believe that social media discourse is negative and divisive. Here we show how this negativity can evolve even when users are not motivated to be negative. We propose that social media users seek to differentiate themselves from other users, and it is easier to differentiate oneself through negativity than positivity because negative information is more heterogeneous and counternormative than positive information. This makes users increasingly likely to post negative comments as a conversation unfolds and it becomes more challenging to make unique contributions. Analyzing 2.05 billion comments from 2,150 Reddit communities shows that comments become more negative over time, both within threads and community histories. This trend toward negativity is mediated by the semantic uniqueness of comments, suggesting that it arises from users differentiating themselves. This trend is strongest when initial dialogue is positive, making negative comments highly counternormative. We replicate these patterns in an experiment simulating social media dialogue (n = 3,685). Participants become more negative over time, but only when incentivized to be unique, and especially when dialogue begins positively. These findings suggest that the structure of social media platforms interacts with human motivation to foster a drift toward negativity over time in online discourse.

Memory is essential for well-being, identity, belief formation, and social cognition. While emotionally intense experiences are better remembered, it remains unclear whether emotions affect central (i.e., core elements) or peripheral (i.e., incidental or less critical) information similarly. Prior research has been limited by the lack of standardized and automated tools for assessing memory content at scale. This study utilizes large language models (LLMs) to quantify both emotional arousal in narrative stimuli and memory fidelity for central and peripheral information in verbal free recall of the narrative. We show that emotion enhances the recall of central information while diminishing recall of peripheral information. Our work introduces a reproducible and scalable framework that can be used in future studies to systematically examine how emotional intensity shapes narrative memory, with implications for psychological theory, educational assessment, and clinical evaluation.

Humans reflect on past memories to make sense of an ongoing event. Past work has shown that people retrieve causally related past events during comprehension, but the exact process by which this causal inference occurs remains elusive. Here, we employed a recurrent neural network augmented with an episodic memory buffer to examine how memories are retrieved and integrated based on causal relationships between events. The model was trained to predict upcoming scenes as it watched a television episode. At every time step, the model transformed the current scene into two distinct representations—”value” representing memory content and “key” representing memory address—both of which were stored as episodic memory. The model learned to retrieve selective past values by applying self-attention over stored keys, and it integrated these retrieved values with the current scene representation to predict an upcoming scene. By separating representations used for encoding from those used for retrieval, the model learned to retrieve memories in ways that go beyond simple pattern similarity. In turn, the model represented causally related events with similar patterns beyond perceptual or semantic similarities, suggesting that it organized event representations based on latent causal structure. Memories retrieved by the model were similar to those retrieved by human participants who watched the same television episode. The model also exhibited hippocampus-like pattern separation and pattern completion, and its representational structure aligned more closely with human fMRI data than a comparison model without an episodic memory buffer. These findings suggest that the model captures the way humans represent events and retrieve memories based on causal relationships. Together, this work proposes a key-value episodic memory system as a candidate computational mechanism for how humans retrieve causally related memories to comprehend naturalistic events. ### Competing Interest Statement The authors have declared no competing interest. McDonnell Center for Systems Neuroscience and the McDonnell Center for Cellular and Molecular Neurobiology at Washington University in St. Louis

Human affective experience varies along the dimensions of valence (positivity or negativity) and arousal (high or low activation). It remains unclear how these dimensions are represented in the brain and whether the representations are shared across different individuals and diverse situational contexts. In this study, we first utilized two publicly available functional MRI datasets of participants watching movies to build predictive models of moment-to-moment emotional arousal and valence from dynamic functional brain connectivity. We tested the models by predicting emotional arousal and valence both within and across datasets. Our results revealed a generalizable arousal representation characterized by the interactions between multiple large-scale functional networks. The arousal representation generalized to two additional movie-watching datasets with different participants viewing different movies. In contrast, we did not find evidence of a generalizable valence representation. Taken together, our findings reveal a generalizable representation of emotional arousal embedded in patterns of dynamic functional connectivity, suggesting a common underlying neural signature of emotional arousal across individuals and situational contexts. We have made our model and analysis scripts publicly available to facilitate its use by other researchers in decoding moment-to-moment emotional arousal in novel datasets, providing a new tool to probe affective experience using fMRI.

In the attention economy of social media, moralized commentary spreads widely, 15 attracts engagement, and is amplified by algorithms. Yet little is known about temporal trends in moralization—specifically, whether moralization has steadily increased, a trend that could deepen divisions and fuel polarization. Using natural language processing, we analyzed 9.7M Twitter/X posts and found a sharp rise in moralized language from 2013-2021. This trend generalized to Reddit (2.1B comments) and outpaced changes in two traditional media corpora 20 (4.9M and 115K texts). Two processes explained this moralizing shift: (1) within-user increases in moral language over time, and (2) selection effects, whereby highly moralized users became more active while less moralized users disengaged. These findings reveal how user dynamics can create highly moralized discourse; understanding this process is crucial for fostering healthier digital ecosystems.

Emotional events tend to be vividly remembered. While growing evidence suggests that emotions have their basis in brain-wide network interactions, it is unclear whether and how these whole-brain dynamics contribute to memory encoding. Here we combined functional MRI, graph theory, text analyses and pupillometry in a naturalistic context where participants recalled complex narratives in their own words. Across three independent datasets, emotionally arousing moments during narrative perception were associated with an integrated brain state characterized by increased cohesion across functional modules, which in turn predicted the fidelity of subsequent recall. Network integration mediated the influence of emotional arousal on recall fidelity, with consistent within- and between-network interactions supporting the mediation across datasets. Together, these results suggest that emotional arousal enhances memory encoding via strengthening functional integration across brain networks. Our findings advance a cross-level understanding of emotional memories that bridges large-scale brain network dynamics, affective states and ongoing cognition.

The Algonauts 2025 Challenge called on the community to develop encoding models that predict whole-brain fMRI responses to naturalistic multimodal movies. In this submission, we propose a sequence-to-sequence Transformer that autoregressively predicts fMRI activity from visual, auditory, and language inputs. Stimulus features were extracted using pretrained models including VideoMAE, HuBERT, Qwen, and BridgeTower. The decoder integrates information from prior brain states, current stimuli, and episode-level summaries via dual cross-attention mechanisms that attend to both perceptual information extracted from the stimulus as well as narrative information provided by high-level summaries of narrative content. One core innovation of our approach is the use of sequences of multimodal context to predict sequences of brain activity, enabling the model to capture long-range temporal structure in both stimuli and neural responses. Another is the combination of a shared encoder with partial subject-specific decoder, which leverages common structure across subjects while accounting for individual variability. Our model achieves strong performance on both in-distribution and out-of-distribution data, demonstrating the effectiveness of temporally-aware, multimodal sequence modeling for brain activity prediction. The code is available at https://github.com/Angelneer926/Algonauts_challenge.

Do goals, beliefs, and desires affect visual experience? This question has long been controversial in cognitive science. There exists extensive literature documenting motivational effects on perceptual reports, but these findings could reflect biases in what people report seeing rather than what they see. Here, we propose that examining the underlying neurocomputational processes can provide new perspectives on this longstanding debate. We review evidence suggesting that motivation biases both perception and action, but does so via distinct neural systems: amygdala and locus coeruleus (LC)-norepinephrine (NE) activity enhances sensory representations for desirable stimuli, while striatal dopamine biases action selection toward goal-congruent actions. The neurocomputational approach provides a framework to advance a mechanistic understanding of motivated seeing and how these biases are shaped by context.

Functional near-infrared spectroscopy (fNIRS) offers a portable, cost-effective alternative to functional magnetic resonance imaging (fMRI) for noninvasively measuring neural activity. However, fNIRS measurements are limited to cortical regions near the scalp, missing important medial and deeper brain areas. We introduce a predictive model that maps prefrontal fNIRS signals to whole-brain fMRI activity during movie-watching. By aligning neural responses to a common audiovisual stimulus, our approach leverages shared dynamics across imaging modalities to map fNIRS signals to broader neural activity patterns. We scanned participants with fNIRS and utilized a publicly available fMRI dataset of participants watching the same TV episode. The model was trained on the first half of the episode and tested on a held-out participant watching the second half to assess cross-individual and cross-stimulus generalizability. The model significantly predicted fMRI time courses in 66 out of 122 brain regions, including areas otherwise inaccessible to fNIRS. It also replicated intersubject functional connectivity patterns and retained semantic information about the movie content. The model generalized to an independent dataset from a different TV series, suggesting it captures robust cross-modal mappings across stimuli. Our publicly available models enable researchers to infer broader neural dynamics from localized fNIRS data during naturalistic tasks.

Perceptual judgments are often influenced by goals and preferences, resulting in biased judgments that deviate from objective reality. When presented with ambiguous images, observers are biased to report seeing images associated with rewards. However, it remains unclear whether this is driven by a bias toward stimuli that are desirable or stimuli that are motivationally salient. As rewards are both desirable and motivationally salient, these effects are not easily dissociated in a reward context. This study investigates the effects of desirability and motivational salience on perceptual judgments in an aversive context involving financial losses. Across two experiments conducted between 2023 and 2024, participants completed a visual categorization task where ambiguous stimuli were associated with a large financial loss. Participants' perceptual judgments were biased away from stimuli associated with the loss, indicating a desirability bias. Drift diffusion model analyses revealed that this bias was due to a shift in the starting point of evidence accumulation, such that participants required more evidence to commit to a response associated with an undesirable outcome. The bias in starting point correlated with individual differences in punishment sensitivity but not reward sensitivity, highlighting how individual traits shape motivational effects on perceptual decisions. Results replicated across an in-lab sample and a larger online sample. Altogether, our study provides robust evidence of a desirability bias in perceptual decisions involving financial losses, identifying both the computational mechanisms and trait-level differences that influence how people decide what they see when faced with the prospect of undesirable outcomes. (PsycInfo Database Record (c) 2025 APA, all rights reserved).