The Language of Creativity: Evidence from Humans and Large Language Models.

Recent developments in computerized scoring via semantic distance have provided automated assessments of verbal creativity. Here, we extend past work, applying computational linguistic approaches to characterize salient features of creative text. We hypothesize that, in addition to semantic diversity, the degree to which a story includes perceptual details, thus transporting the reader to another time and place, would be predictive of creativity. Additionally, we explore the use of generative language models to supplement human data collection and examine the extent to which machine-generated stories can mimic human creativity. We collect 600 short stories from human participants and GPT-3, subsequently randomized and assessed on their creative quality. Results indicate that the presence of perceptual details, in conjunction with semantic diversity, is highly predictive of creativity. These results were replicated in an independent sample of stories (n = 120) generated by GPT-4. We do not observe a significant difference between human and AI-generated stories in terms of creativity ratings, and we also observe positive correlations between human and AI assessments of creativity. Implications and future directions are discussed.

Eye movements reinstate remembered locations during episodic simulation.

Imagining the future, like recalling the past, relies on the ability to retrieve and imagine a spatial context. Research suggests that eye movements support this process by reactivating spatial contextual details from memory, a process termed gaze reinstatement. While gaze reinstatement has been linked to successful memory retrieval, it remains unclear whether it supports the related process of future simulation. In the present study, we recorded both eye movements and audio while participants described familiar locations from memory and subsequently imagined future events occurring in those locations while either freely moving their eyes or maintaining central fixation. Restricting viewing during simulation significantly reduced self-reported vividness ratings, supporting a critical role for eye movements in simulation. When viewing was unrestricted, participants spontaneously reinstated gaze patterns specific to the simulated location, replicating findings of gaze reinstatement during memory retrieval. Finally, gaze-based location reinstatement was predictive of simulation success, indexed by the number of internal (episodic) details produced, with both measures peaking early and co-varying over time. Together, these findings suggest that the same oculomotor processes that support episodic memory retrieval - that is, gaze-based reinstatement of spatial context - also support episodic simulation.

Peering into the future: Eye movements predict neural repetition effects during episodic simulation.

Imagining future scenarios involves recombining different elements of past experiences into a coherent event, a process broadly supported by the brain's default network. Prior work suggests that distinct brain regions may contribute to the inclusion of different simulation features. Here we examine how activity in these brain regions relates to the vividness of future simulations. Thirty-four healthy young adults imagined future events with familiar people and locations in a two-part study involving a repetition suppression paradigm. First, participants imagined events while their eyes were tracked during a behavioral session. Immediately after, participants imagined events during MRI scanning. The events to be imagined were manipulated such that some were identical to those imagined in the behavioral session while others involved new locations, new people, or both. In this way, we could examine how self-report ratings and eye movements predict brain activity during simulation along with specific simulation features. Vividness ratings were negatively correlated with eye movements, in contrast to an often-observed positive relationship with past recollection. Moreover, fewer eye movements predicted greater involvement of the hippocampus during simulation, an effect specific to location features. Our findings suggest that eye movements may facilitate scene construction for future thinking, lending support to frameworks that spatial information forms the foundation of episodic simulation.

Endel Tulving (1927-2023).

This article presents an obituary for Endel Tulving. Tulving's educational and professional careers are summarized. His work in the field of human memory is detailed. It is noted that Tulving's look at the field of verbal learning in the late 1950s persuaded him that the dominant associative tradition missed many important aspects of human memory. His research found that at the time of retrieval, memory for the original event may be successfully reinstated only by contextual cues that interact in a complementary fashion with the specifically encoded memory trace, a process that Tulving referred to as "synergistic ecphory". He is also known for his work on memory systems. In his book, Elements of Episodic Memory published in 1983, Tulving proposed that memory for experienced events, episodic memory, should be distinguished from general knowledge of the world, semantic memory, and from procedural memory, the learned ability to perform such skilled procedures as riding a bicycle or playing a musical instrument. He also proposed an evolutionary framework for these different but related systems, suggesting that simple animals show only procedural memory, more complex animals are consciously aware of their knowledge of the world, but only humans possess episodic memory-the ability to use "mental time travel" to consciously recreate past experiences and to imagine possible future events. Although known initially for his purely cognitive behavioral research, during the 1980s and 1990s, Tulving increasingly incorporated neuropsychological and neuroimaging approaches into his work. (PsycInfo Database Record (c) 2024 APA, all rights reserved).

Automated scoring of the autobiographical interview with natural language processing.

The autobiographical interview has been used in more than 200 studies to assess the content of autobiographical memories. In a typical experiment, participants recall memories, which are then scored manually for internal details (episodic details from the central event) and external details (largely non-episodic details). Scoring these narratives requires a significant amount of time. As a result, large studies with this procedure are often impractical, and even conducting small studies is time-consuming. To reduce scoring burden and enable larger studies, we developed an approach to automatically score responses with natural language processing. We fine-tuned an existing language model (distilBERT) to identify the amount of internal and external content in each sentence. These predictions were aggregated to obtain internal and external content estimates for each narrative. We evaluated our model by comparing manual scores with automated scores in five datasets. We found that our model performed well across datasets. In four datasets, we found a strong correlation between internal detail counts and the amount of predicted internal content. In these datasets, manual and automated external scores were also strongly correlated, and we found minimal misclassification of content. In a fifth dataset, our model performed well after additional preprocessing. To make automated scoring available to other researchers, we provide a Colab notebook that is intended to be used without additional coding.

An examination of episodic future thinking in the emergency department among youth experiencing suicidal thoughts and behaviors.

Youth experiencing suicidal thoughts and/or behaviors (STBs) frequently present to emergency departments for acute psychiatric care. These settings offer a transitory yet pivotal opportunity to assess, intervene on, and plan continued care for STBs. This study examined a clinically relevant, understudied aspect of psychological functioning among youth experiencing STBs in the emergency department: episodic future thinking, or the ability to imagine discrete autobiographical future events. A sample of 167 youths (10-17 years) presenting to a pediatric psychiatric emergency department for STBs completed a performance-based measure of episodic future thinking assessing richness in detail and subjective characteristics of imagined future events. STB recurrence was assessed 6 months later. Immediately following a suicide-related crisis, youth demonstrated mixed abilities to imagine their future: they generated some concrete future event details but did not subjectively perceive these events as being very detailed or likely to occur. Older adolescents (i.e., 15-17) generated more episodic details than pre-/younger adolescents (i.e., 10-14), particularly those pertaining to actions or sensory perceptions. There was no evidence linking less detailed episodic future thinking and greater likelihood of STBs following the emergency department visit; instead, hopelessness was a more robust risk factor. Findings underscore the importance and clinical utility of better understanding the psychological state of youth during or immediately following a suicide-related crisis. In particular, assessing youths' future thinking abilities in the emergency department may directly inform approaches to acute care delivery.

Grandiose narcissism influences the phenomenology of remembered past and imagined future events.

Little empirical work has examined future thinking in narcissistic grandiosity. We here extend prior work finding that people scoring high in grandiosity have self-bolstering tendencies in remembering past events, and we consider whether these tendencies extend to imagining future events. Across an initial study (N = 112) and replication (N = 169), participants wrote about remembered past events and imagined future events in which they embodied or would embody either positive or negative traits. Participants then rated those events on several subjective measures. We find that people scoring higher in grandiosity remember past events in which they embody positive traits with greater detail and ease than past events in which they embody negative traits. These same effects persist when people scoring high in grandiosity imagine possible events in their future. Those scoring higher in grandiosity endorse thinking about positive events in their past and future more frequently than negative events, and they judge positive future events as more plausible than negative future events. These tendencies did not extend to objective detail provided in their written narratives about these events. Taken together, these findings demonstrate that grandiosity is associated with self-bolstering tendencies in both remembering the past and imagining the future.

Retrieval fluency inflates perceived preparation for difficult problems.

When faced with a difficult problem, people often rely on past experiences. While remembering clearly helps us reach solutions, can retrieval also lead to misperceptions of our abilities? In three experiments, participants encountered "worst case scenarios" they likely had never experienced and that would be difficult to navigate without extensive training (e.g., bitten by snake). Learning brief tips improved problem-solving performance later, but retrieval increased feelings of preparation by an even larger margin. This gap occurred regardless of whether people thought that tips came from an expert or another participant in the study, and it did not reflect mere familiarity with the problems themselves. Instead, our results suggest that the ease experienced while remembering, or retrieval fluency, inflated feelings of preparation.

Creativity at rest: Exploring functional network connectivity of creative experts.

The neuroscience of creativity seeks to disentangle the complex brain processes that underpin the generation of novel ideas. Neuroimaging studies of functional connectivity, particularly functional magnetic resonance imaging (fMRI), have revealed individual differences in brain network organization associated with creative ability; however, much of the extant research is limited to laboratory-based divergent thinking measures. To overcome these limitations, we compare functional brain connectivity in a cohort of creative experts (n = 27) and controls (n = 26) and examine links with creative behavior. First, we replicate prior findings showing reduced connectivity in visual cortex related to higher creative performance. Second, we examine whether this result is driven by integrated or segregated connectivity. Third, we examine associations between functional connectivity and vivid distal simulation separately in creative experts and controls. In accordance with past work, our results show reduced connectivity to the primary visual cortex in creative experts at rest. Additionally, we observe a negative association between distal simulation vividness and connectivity to the lateral visual cortex in creative experts. Taken together, these results highlight connectivity profiles of highly creative people and suggest that creative thinking may be related to, though not fully redundant with, the ability to vividly imagine the future.

Episodic simulation of helping behavior in younger and older adults during the COVID-19 pandemic.

Imagining helping a person in need increases one's willingness to help beyond levels evoked by passively reading the same stories. We examined whether episodic simulation can increase younger and older adults' willingness to help in novel scenarios posed by the COVID-19 pandemic. Across 3 studies we demonstrate that episodic simulation of helping behavior increases younger and older adults' willingness to help during both everyday and COVID-related scenarios. Moreover, we show that imagining helping increases emotional concern, scene imagery, and theory of mind, which in turn relate to increased willingness to help. Studies 2 and 3 also showed that people produce more internal, episodic-like details when imagining everyday compared to COVID-related scenarios, suggesting that people are less able to draw on prior experiences when simulating such novel events. These findings suggest that encouraging engagement with stories of people in need by imagining helping can increase willingness to help during the pandemic.

A language of episodic thought?

We propose that episodic thought (i.e., episodic memory and imagination) is a domain where the language-of-thought hypothesis (LoTH) could be fruitfully applied. On the one hand, LoTH could explain the structure of what is encoded into and retrieved from long-term memory. On the other, LoTH can help make sense of how episodic contents come to play such a large variety of different cognitive roles after they have been retrieved.

Intentionality of Self-Generated Thought: Contributions of Mind Wandering to Creativity.

Studies suggest that internally oriented cognitive processes are central to creativity. Here, we distinguish between intentional and unintentional forms of mind wandering and explore their behavioral and neural correlates. We used a sample of 155 healthy adults from the mind-brain-body dataset, all of whom completed resting-state fMRI scans and trait-level measures of mind wandering. We analyzed intentional and unintentional mind wandering tendencies using self-report measures. Next, we explored the relationship between mind wandering tendencies and creativity, as measured by a divergent thinking task. Finally, we describe patterns of resting-state network connectivity associated with mind wandering, using graph theory analysis. At the behavioral level, results showed a significant positive association between creativity and both intentional and unintentional mind wandering. Neuroimaging analysis revealed higher weighted degree connectivity associated with both forms of mind wandering, implicating core regions of the default network and the left temporal pole. We observed topological connectivity differences within the default network: intentional mind wandering was associated with degree connectivity in posterior regions, whereas unintentional mind wandering showed greater involvement of prefrontal areas. Overall, the findings highlight patterns of resting-state network connectivity associated with intentional and unintentional mind wandering, and provide novel evidence of a link between mind wandering and creativity.

Bias and constructive processes in a self-memory system.

Martin Conway's influential theorising about the self-memory system (Conway, M. A., & Pleydell-Pearce, C. W. (2000). The construction of autobiographical memories in the self-memory system. Psychological Review, 107(2), 261-288) illuminated how the "working self" influences the construction of autobiographical memories. Moreover, his constructive view of self and memory is compatible with the occurrence of various kinds of errors and distortions in remembering. Here we consider one of the "seven sins" of memory Schacter, D. L. (2021). The seven sins of memory updated edition: How the mind forgets and remembers. Houghton Mifflin Harcourt) that we believe is most closely related to the operation of Conway's self-memory system: bias, which refers to the role of current knowledge, beliefs, and feelings in shaping and sometimes distorting memories for past experiences and attitudes. More specifically, we discuss recent research on three forms of bias - consistency, self-enhancing, and positivity biases - that illuminate their role in influencing how people remember the past and also imagine the future. We consider both theoretical and applied aspects of these biases and, consistent with Conway's perspective, argue that despite sometimes contributing to inaccuracies, bias also serves adaptive functions.

Semantic Memory and Creativity: The Costs and Benefits of Semantic Memory Structure in Generating Original Ideas.

Despite its theoretical importance, little is known about how semantic memory structure facilitates and constrains creative idea production. We examine whether the semantic richness of a concept has both benefits and costs to creative idea production. Specifically, we tested whether cue set-size-an index of semantic richness reflecting the average number of elements associated with a given concept-impacts the quantity (fluency) and quality (originality) of responses generated during the alternate uses task (AUT). Across four studies, we show that low-association, sparse, AUT cues benefit originality at the cost of fluency compared to high-association, rich, AUT cues. Furthermore, we found an interaction with individual differences in fluid intelligence in the low-association AUT cues, suggesting that constraints of sparse semantic knowledge can be overcome with top-down intervention. The findings indicate that semantic richness differentially impacts the quality and quantity of generated ideas, and that cognitive control processes can facilitate idea production when conceptual knowledge is limited.

Common and distinct correlates of construction and elaboration of episodic-autobiographical memory: An ALE meta-analysis.

The recollection of episodic-autobiographical memories (EAMs) entails a complex temporal dynamic, from initial "construction" to subsequent "elaboration" of memories. While there is consensus that EAM retrieval involves a distributed network of brain regions, it is still largely debated which regions specifically contribute to EAM construction and/or elaboration. To clarify this issue, we conducted an Activation Likelihood Estimation (ALE) meta-analysis based on the Preferred Reporting Items for Systematic-Reviews and Meta-Analyses (PRISMA) method. We found common recruitment of the left hippocampus and posterior cingulate cortex (PCC) during both phases. Additionally, EAM construction led to activations in the ventromedial prefrontal cortex, left angular gyrus (AG), right hippocampus, and precuneus, while the right inferior frontal gyrus was activated by EAM elaboration. Although most of these regions are distributed over the default mode network, the current findings highlight a differential contribution according to early (midline regions, left/right hippocampus, and left AG) versus later (left hippocampus, and PCC) recollection. Overall, these findings contribute to clarify the neural correlates that support the temporal dynamics of EAM recollection.

Mnemicity: A Cognitive Gadget?

Episodic representations can be entertained either as "remembered" or "imagined"-as outcomes of experience or as simulations of such experience. Here, we argue that this feature is the product of a dedicated cognitive function: the metacognitive capacity to determine the mnemicity of mental event simulations. We argue that mnemicity attribution should be distinguished from other metacognitive operations (such as reality monitoring) and propose that this attribution is a "cognitive gadget"-a distinctively human ability made possible by cultural learning. Cultural learning is a type of social learning in which traits are inherited through social interaction. In the case of mnemicity, one culturally learns to discriminate metacognitive "feelings of remembering" from other perceptual, emotional, action-related, and metacognitive feelings; to interpret feelings of remembering as indicators of memory rather than imagination; and to broadcast the interpreted feelings in culture- and context-specific ways, such as "I was there" or "I witnessed it myself." We review evidence from the literature on memory development and scaffolding, metacognitive learning and teaching, as well as cross-cultural psychology in support of this view before pointing out various open questions about the nature and development of mnemicity highlighted by our account.

Constructive episodic retrieval processes underlying memory distortion contribute to creative thinking and everyday problem solving.

Constructive episodic retrieval processes play an adaptive role in supporting divergent thinking (i.e., creatively combining diverse bits of information) and means-end problem solving (i.e., generating steps to solve a social problem). However, the constructive nature of episodic memory that supports these adaptive functions also leads to memory error. In three experiments we aimed to identify a direct link between divergent thinking and means-end problem solving - as assessed in the Alternative Uses Task (AUT) and Means-End Problem Solving (MEPS) task - with the generation of false memories in the Deese-Roediger-McDermott paradigm. In Experiment 1, we replicated prior findings where false memory was positively correlated with performance on the AUT, and also showed for the first time that increased performance in the MEPS task is associated with increased false recall. In Experiment 2, we demonstrated that the link between false recall and performance on the MEPS task did not extend to other forms of problem solving, as assessed with the Everyday Descriptions Task (EDT). In Experiment 3, we showed that when the EDT was preceded by the MEPS task in an attempt to influence participants to engage in a similar episodic-problem solving strategy, performance in both tasks was correlated with false memory. These findings provide evidence for a direct link between the adaptive benefits of constructive episodic processes, in the form of enhanced divergent creative thinking and problem solving, and costs, in the form of increased memory error.

Transcranial magnetic stimulation to the angular gyrus modulates the temporal dynamics of the hippocampus and entorhinal cortex.

Transcranial magnetic stimulation (TMS) delivered to the angular gyrus (AG) affects hippocampal function and associated behaviors (Thakral PP, Madore KP, Kalinowski SE, Schacter DL. Modulation of hippocampal brain networks produces changes in episodic simulation and divergent thinking. 2020a. Proc Natl Acad Sci U S A. 117:12729-12740). Here, we examine if functional magnetic resonance imaging (fMRI)-guided TMS disrupts the gradient organization of temporal signal properties, known as the temporal organization, in the hippocampus (HPC) and entorhinal cortex (ERC). For each of 2 TMS sessions, TMS was applied to either a control site (vertex) or to a left AG target region (N = 18; 14 females). Behavioral measures were then administered, and resting-state scans were acquired. Temporal dynamics were measured by tracking change in the fMRI signal (i) "within" single voxels over time, termed single-voxel autocorrelation and (ii) "between" different voxels over time, termed intervoxel similarity. TMS reduced AG connectivity with the hippocampal target and induced more rapid shifting of activity in single voxels between successive time points, lowering the single-voxel autocorrelation, within the left anteromedial HPC and posteromedial ERC. Intervoxel similarity was only marginally affected by TMS. Our findings suggest that hippocampal-targeted TMS disrupts the functional properties of the target site along the anterior/posterior axis. Further studies should examine the consequences of altering the temporal dynamics of these medial temporal areas to the successful processing of episodic information under task demand.