Brain activity in using heuristic prototype to solve insightful problems

Hierarchical Prototype Networks for Continual Graph Representation Learning

Despite significant advances in graph representation learning, little attention has been paid to the more practical continual learning scenario in which new categories of nodes (e.g., new research areas in citation networks, or new types of products in co-purchasing networks) and their associated edges are continuously emerging, causing catastrophic forgetting on previous categories. Existing methods either ignore the rich topological information or sacrifice plasticity for stability. To this end, we present Hierarchical Prototype Networks (HPNs) which extract different levels of abstract knowledge in the form of prototypes to represent the continuously expanded graphs. Specifically, we first leverage a set of Atomic Feature Extractors (AFEs) to encode both the elemental attribute information and the topological structure of the target node. Next, we develop HPNs to adaptively select relevant AFEs and represent each node with three levels of prototypes. In this way, whenever a new category of nodes is given, only the relevant AFEs and prototypes at each level will be activated and refined, while others remain uninterrupted to maintain the performance over existing nodes. Theoretically, we first demonstrate that the memory consumption of HPNs is bounded regardless of how many tasks are encountered. Then, we prove that under mild constraints, learning new tasks will not alter the prototypes matched to previous data, thereby eliminating the forgetting problem. The theoretical results are supported by experiments on five datasets, showing that HPNs not only outperform state-of-the-art baseline techniques but also consume relatively less memory. Code and datasets are available at https://github.com/QueuQ/HPNs.

Unconscious processing of prototype heuristics in scientific innovation problem-solving

Previously published studies on the effect of how different levels of unconsciousness (UC) and different loads of executive functions (EFs) affect insight problem solving are inconsistent. In a set of three experiments, we used scientific innovation problems (SIP) as insight metrics and distractor tasks to induce UC. Experiment 1 confirmed that, compared with conscious processing, unconscious processing is more conducive to obtaining prototype heuristics for correctly solving scientific innovation problems creatively. Furthermore, Experiment 2 found that different levels of unconscious processing, which were induced by different distractor tasks, made a different impact on high or low difficulty creative problem solving. Experiment 3 indicated that unconscious processing could improve prototype activation and the ability to use key heuristics information in prototype heuristics processing by improving working memory, inhibitory control, and shifting ability of EFs. Overall, the present results provide additional evidence for the role of consciousness levels in insight problem solving.

Uncovering neural distinctions and commodities between two creativity subsets: A meta-analysis of fMRI studies in divergent thinking and insight using activation likelihood estimation

The dual‐process theory that two different systems of thought coexist in creative thinking has attracted considerable attention. In the field of creative thinking, divergent thinking (DT) is the ability to produce multiple solutions to open‐ended problems in a short time. It is mainly considered an associative and fast process. Meanwhile, insight, the new and unexpected comprehension of close‐ended problems, is frequently marked as a deliberate and time‐consuming thinking process requiring concentrated effort. Previous research has been dedicated to revealing their separate neural mechanisms, while few studies have compared their differences and similarities at the brain level. Therefore, the current study applied Activation Likelihood Estimation to decipher common and distinctive neural pathways that potentially underlie DT and insight. We selected 27 DT studies and 30 insight studies for retrospective meta‐analyses. Initially, two single analyses with follow‐up contrast and conjunction analyses were performed. The single analyses showed that DT mainly involved the inferior parietal lobe (IPL), cuneus, and middle frontal gyrus (MFG), while the precentral gyrus, inferior frontal gyrus (IFG), parahippocampal gyrus (PG), amygdala (AMG), and superior parietal lobe were engaged in insight. Compared to insight, DT mainly led to greater activation in the IPL, the crucial part of the default mode network. However, insight caused more significant activation in regions related to executive control functions and emotional responses, such as the IFG, MFG, PG, and AMG. Notably, the conjunction analysis detected no overlapped areas between DT and insight. These neural findings implicate that various neurocognitive circuits may support DT and insight.

Creative problem solving in knowledge-rich contexts

Creative problem solving (CPS) in real-world contexts often relies on reorganization of existing knowledge to serve new, problem-relevant functions. However, classic creativity paradigms that minimize knowledge content are generally used to investigate creativity, including CPS. We argue that CPS research should expand consideration of knowledge-rich problem contexts, both in novices and experts within specific domains. In particular, paradigms focusing on creative analogical transfer of knowledge may reflect CPS skills that are applicable to real-world problem solving. Such paradigms have begun to provide process-level insights into cognitive and neural characteristics of knowledge-rich CPS and point to multiple avenues for fruitfully expanding inquiry into the role of crystalized knowledge in creativity.

Mapping the artistic brain: Common and distinct neural activations associated with musical, drawing, and literary creativity

Whether creativity is a domain‐general or domain‐specific ability has been a topic of intense speculation. Although previous studies have examined domain‐specific mechanisms of creative performance, little is known about commonalities and distinctions in neural correlates across different domains. We applied activation likelihood estimation (ALE) meta‐analysis to identify the brain activation of domain‐mechanisms by synthesizing functional neuroimaging studies across three forms of artistic creativity: music improvisation, drawing, and literary creativity. ALE meta‐analysis yielded a domain‐general pattern across three artistic forms, with overlapping clusters in the presupplementary motor area (pre‐SMA), left dorsolateral prefrontal cortex, and right inferior frontal gyrus (IFG). Regarding domain‐specificity, musical creativity was associated with recruitment of the SMA‐proper, bilateral IFG, left precentral gyrus, and left middle frontal gyrus (MFG) compared to the other two artistic forms; drawing creativity recruited the left fusiform gyrus, left precuneus, right parahippocampal gyrus, and right MFG compared to musical creativity; and literary creativity recruited the left angular gyrus and right lingual gyrus compared to musical creativity. Contrasting drawing and literary creativity revealed no significant differences in neural activation, suggesting that these domains may rely on a common neurocognitive system. Overall, these findings reveal a central, domain‐general system for artistic creativity, but with each domain relying to some degree on domain‐specific neural circuits.

Critical thinking and regional gray matter volume interact to predict representation connection in scientific problem solving

Representation connection (RC) is a stable ability that significantly predicts the accuracy of scientific innovation problem solving while critical thinking has been strongly related to problem solving. However, the neural mechanisms underlying this relationship have not been assessed. Using voxel-based morphometry (VBM) and scientific innovation problem solving materials, we investigated the correlation between RC and regional gray matter volume (rGMV) in healthy young participants. We found that RC was positively correlated with rGMV in the right superior temporal gyrus (STG) and in a cluster in the left medial frontal gyrus (MFG). These results indicate that increased rGMV in the right STG may lead to the ability to overcome misdirection more easily, which may result in better semantic integration of the "certain construction" of heuristic prototypes. Increased rGMV in the left MFG may be associated with forming novel associations and retrieving matched unsolved technical problems from memory. Further analysis revealed that the interaction between critical thinking and rGMV predicted RC in insightful problem solving, and found that higher rGMV was correlated with higher RC in participants with lower cognitive maturity, but not in participants with higher cognitive maturity. These findings suggest that rGMV could interact with cognitive maturity to modulate RC in insightful problem solving.

Cognitive conflict and restructuring: The neural basis of two core components of insight

Sometimes, the solution to a difficult problem simply pops into mind. Such a moment of sudden comprehension is known as "insight". This fundamental cognitive process is crucial for problem solving, creativity and innovation, yet its true nature remains elusive, despite one century of psychological research. Typically, insight is investigated by using spatial puzzles or verbal riddles. Broadening the traditional approach, we propose to tackle this question by presenting magic tricks to participants and asking them to find out the secret method used by the magician. Combining this approach with cueing in an fMRI experiment, we were able to break down the insight process into two underlying components: cognitive conflict and restructuring. During cognitive conflict, problem solvers identify incongruent information that does not match their current mental representation. In a second step this information is restructured, thereby allowing them to correctly determine how the magic trick was done. We manipulated the occurrence of cognitive conflict by presenting two types of cues that lead participants to either maintain their perceptual belief (congruent cue) or to change their perceptual belief (incongruent cue) for the mechanism behind the magic trick. We found that partially overlapping but distinct networks of brain activity were recruited for cognitive conflict and restructuring. Posterior, predominantly visual brain activity during cognitive conflict reflected processes related to prediction error, attention to the relevant cue-specific sensory domain, and the default brain state. Restructuring on the other hand, showed a highly distributed pattern of brain activity in regions of the default mode, executive control networks, and salience networks. The angular gyrus and middle temporal gyrus were active in both cognitive conflict and restructuring, suggesting that these regions are important throughout the insight problem solving process. We believe this type of approach towards understanding insight will give lead to a better understanding of this complex process and the specific role that different brain regions play in creative thought.

Real World Problem-Solving

Real world problem-solving (RWPS) is what we do every day. It requires flexibility, resilience, resourcefulness, and a certain degree of creativity. A crucial feature of RWPS is that it involves continuous interaction with the environment during the problem-solving process. In this process, the environment can be seen as not only a source of inspiration for new ideas but also as a tool to facilitate creative thinking. The cognitive neuroscience literature in creativity and problem-solving is extensive, but it has largely focused on neural networks that are active when subjects are not focused on the outside world, i.e., not using their environment. In this paper, I attempt to combine the relevant literature on creativity and problem-solving with the scattered and nascent work in perceptually-driven learning from the environment. I present my synthesis as a potential new theory for real world problem-solving and map out its hypothesized neural basis. I outline some testable predictions made by the model and provide some considerations and ideas for experimental paradigms that could be used to evaluate the model more thoroughly.

Neural correlates of novelty and appropriateness processing in externally induced constraint relaxation

Novelty and appropriateness are considered the two fundamental features of creative thinking, including insight problem solving, which can be performed through chunk decomposition and constraint relaxation. Based on a previous study that separated the neural bases of novelty and appropriateness in chunk decomposition, in this study, we used event-related functional magnetic resonance imaging (fMRI) to further dissociate these mechanisms in constraint relaxation. Participants were guided to mentally represent the method of problem solving according to the externally provided solutions that were elaborately prepared in advance and systematically varied in their novelty and appropriateness for the given problem situation. The results showed that novelty processing was completed by the temporoparietal junction (TPJ) and regions in the executive system (dorsolateral prefrontal cortex [DLPFC]), whereas appropriateness processing was completed by the TPJ and regions in the episodic memory (hippocampus), emotion (amygdala), and reward systems (orbitofrontal cortex [OFC]). These results likely indicate that appropriateness processing can result in a more memorable and richer experience than novelty processing in constraint relaxation. The shared and distinct neural mechanisms of the features of novelty and appropriateness in constraint relaxation are discussed, enriching the representation of the change theory of insight.

The Structural and Functional Organization of Cognition

This article proposes that what have been historically and contemporarily defined as different domains of human cognition are served by one of four functionally- and structurally-distinct areas of the prefrontal cortex (PFC). Their contributions to human intelligence are as follows: (a) BA9, enables our emotional intelligence, engaging the psychosocial domain; (b) BA47, enables our practical intelligence, engaging the material domain; (c) BA46 (or BA46-9/46), enables our abstract intelligence, engaging the hypothetical domain; and (d) BA10, enables our temporal intelligence, engaging in planning within any of the other three domains. Given their unique contribution to human cognition, it is proposed that these areas be called the, social (BA9), material (BA47), abstract (BA46-9/46) and temporal (BA10) mind. The evidence that BA47 participates strongly in verbal and gestural communication suggests that language evolved primarily as a consequence of the extreme selective pressure for practicality; an observation supported by the functional connectivity between BA47 and orbital areas that negatively reinforce lying. It is further proposed that the abstract mind (BA46-9/46) is the primary seat of metacognition charged with creating adaptive behavioral strategies by generating higher-order concepts (hypotheses) from lower-order concepts originating from the other three domains of cognition.

Is creative insight task-specific? A coordinate-based meta-analysis of neuroimaging studies on insightful problem solving

The question of whether creative insight varies across problem types has recently come to the forefront of studies of creative cognition. In the present study, to address the nature of creative insight, the coordinate-based activation likelihood estimation (ALE) technique was utilized to individually conduct three quantitative meta-analyses of neuroimaging experiments that used the compound remote associate (CRA) task, the prototype heuristic (PH) task and the Chinese character chunk decomposition (CCD) task. These tasks were chosen because they are frequently used to uncover the neurocognitive correlates of insight. Our results demonstrated that creative insight reliably activates largely non-overlapping brain regions across task types, with the exception of some shared regions: the CRA task mainly relied on the right parahippocampal gyrus, the superior frontal gyrus and the inferior frontal gyrus; the PH task primarily depended on the right middle occipital gyrus (MOG), the bilateral superior parietal lobule/precuneus, the left inferior parietal lobule, the left lingual gyrus and the left middle frontal gyrus; and the CCD task activated a broad cerebral network consisting of most dorsolateral and medial prefrontal regions, frontoparietal regions and the right MOG. These results provide the first neural evidence of the task dependence of creative insight. The implications of these findings for resolving conflict surrounding the different theories of creative cognition and for defining insight as a set of heterogeneous processes are discussed.

Cultural Evolutionary Perspectives on Creativity and Human Innovation

Cultural traits originate through creative or innovative processes, which might be crucial to understanding how culture evolves and accumulates. However, because of its complexity and apparent subjectivity, creativity has remained largely unexplored as the dynamic underpinning of cultural evolution. Here, we explore the approach to innovation commonly taken in theoretical studies of cultural evolution and discuss its limitations. Drawing insights from cognitive science, psychology, archeology, and even animal behavior, it is possible to generate a formal description of creativity and to incorporate a dynamic theory of creativity into models of cultural evolution. We discuss the implications of such models for our understanding of the archaeological record and the history of hominid culture.

An illustrated heuristic prototype facilitates scientific inventive problem solving: A functional magnetic resonance imaging study

Many scientific inventions (SI) throughout history were inspired by heuristic prototypes (HPs). For instance, an event or piece of knowledge similar to displaced water from a tub inspired Archimedes' principle. However, the neural mechanisms underlying this insightful problem solving are not very clear. Thus, the present study explored the neural correlates used to solve SI problems facilitated by HPs. Each HP had two versions: a literal description with an illustration (LDI) and a literal description with no illustration (LDNI). Thirty-two participants were divided randomly into these two groups. Blood oxygenation level-dependent fMRI contrasts between LDI and LDNI groups were measured. Greater activity in the right middle occipital gyrus (RMOG, BA19), right precentral gyrus (RPCG, BA4), and left middle frontal gyrus (LMFG, BA46) were found within the LDI group as compared to the LDNI group. We discuss these results in terms cognitive functions within these regions related to problem solving and memory retrieval.

Dissociable neural systems underwrite logical reasoning in the context of induced emotions with positive and negative valence

How emotions influence syllogistic reasoning is not well understood. fMRI was employed to investigate the effects of induced positive or negative emotion on syllogistic reasoning. Specifically, on a trial-by-trial basis participants were exposed to a positive, negative, or neutral picture, immediately prior to engagement in a reasoning task. After viewing and rating the valence and intensity of each picture, participants indicated by keypress whether or not the conclusion of the syllogism followed logically from the premises. The content of all syllogisms was neutral, and the influence of belief-bias was controlled for in the study design. Emotion did not affect reasoning performance, although there was a trend in the expected direction based on accuracy rates for the positive (63%) and negative (64%) versus neutral (70%) condition. Nevertheless, exposure to positive and negative pictures led to dissociable patterns of neural activation during reasoning. Therefore, the neural basis of deductive reasoning differs as a function of the valence of the context.