When we appreciate works of art, is the brain faithfully reproducing reality, or is it building personalized visual experiences with a filter? This inquiry is related to the "beholder's share" theory proposed by Nobel laureate in Physiology or Medicine, Eric Kandel. This neuroaesthetic concept will serve as the entry point to analyze the operating mechanism of the human brain during art appreciation. The answer to this question may explain why, several years after the emergence of AI drawing, Ghibli-style illustrations generated by ChatGPT can still go viral.
Image source: https://easy-peasy.ai/ai-image-generator/images/abstract-neuroscience-art-brain-neuron-connections-pre-1912-style
The Beholder's Share,
You Are Not a Passive Art Appreciator
Vienna, in the eyes of contemporary people, is a city of music, where the Baroque architecture of Schönbrunn Palace and the art treasures in the museum district complement each other; those deeply touching masterpieces always stir the most subtle emotional tremors in viewers. A hundred years ago, numerous mathematicians and logicians of the Vienna School attempted to point out through logical positivism that experience and logic are the only reliable sources for understanding the world.
However, when we separate art appreciation from scientific research, our understanding of the essence of cognition often falls into a dilemma: art critics habitually use obscure terminology to describe subjective feelings, like appreciating flowers in the fog, unable to see the whole picture; philosophers, on the other hand, are fixated on abstract theories to explain universal laws, as if scratching an itch through a boot, never touching the core. Eric Kandel, from Vienna, sought to bridge the "two cultures" path proposed by Snow. This Harvard graduate in history and literature, due to his relationship with Anna Kris, a psychoanalyst from the Freudian school, gradually moved beyond the theoretical framework of psychoanalysis and instead explored the biological origins of human perception and memory formation from the study of hippocampal synaptic plasticity.
Eric Kandel's view on explaining art appreciation experiences with neuroscience is detailed in his book "Reductionism in Art and Brain Science." The core concept in the book, the "beholder's share" theory, emphasizes that humans are not passively receiving information when appreciating a painting, but actively participating in visual construction. The visual system transforms the two-dimensional information of the canvas into a three-dimensional spatial mapping in the brain, while simultaneously drawing upon an individual's memories and emotional reserves to imbue the visual elements with unique subjective interpretations. This dual mechanism of neural decoding and psychological projection forms the biological basis of artistic experience.
Neural circuit diagram of "the beholder's share": The neural circuits involved in "the beholder's share" include those processing visual and tactile stimuli, emotions, actions, empathy, and theory of mind. Image source: CHRIS WILCOX, produced/translated by: Cun Yuan
Predictive coding theory in modern neuroscience offers a new perspective on this view, suggesting that perception is essentially a mental landscape actively constructed by the brain. As Anni Seth points out in her paper "From unconscious inference to the beholder’s share: Predictive perception and human experience": "Sensory signals themselves are ambiguous and noisy, and the brain relies on prior knowledge (expectations) to make probabilistic inferences about the source of signals (i.e., Bayesian inference)."
Figure 2, A: Diagram of hierarchical predictive coding across three cortical areas; B: The effect of precision on Bayesian inference and predictive coding. Image source: cell
According to the predictive coding theoretical framework, the innovative creative methods of Impressionist painters align with neuroscientific principles. Monet's fragmented brushstrokes and Cézanne's structural color blocks essentially deliberately create "predictive gaps" in visual information; that is, the viewer's experience determines the meaning of the image, rather than the physical stimulus itself. For example, in Cézanne's "Large Pine and Red Earth," the outlines of the main subject, the pine tree, are intentionally blurred, requiring the viewer to rely on stored memories of tree forms in the hippocampus to complete the outline when processing peripheral vision.
Impressionist works were able to break through the traditional artistic conventions precisely because they challenged traditional visual habits, forcing viewers to rethink how they perceive the world. These paintings do not simply reproduce nature, but rather guide viewers to engage in "predictive perception" through the arrangement of colors and forms, inferring and interpreting the content of the image based on existing experience.
Cézanne's "Large Pine and Red Earth." Image source: artchive.com
Abstract Art is No Longer Difficult to Understand,
Personal Experience Makes It Unique for Everyone
Since the establishment of scientific perspective laws during the Renaissance, Western painting has undergone a paradigm shift from objective representation to subjective expression. After Impressionist painters broke the academic rules of light and shadow, Cubists then deconstructed object forms, until Kandinsky pioneered abstract expressionism, completely stripping away concrete elements. This historical evolution of art provides natural experimental material for neuroaesthetic research.
Dr. Celia Durkin from Columbia University (mentored by Eric Kandel), through behavioral experiments and functional magnetic resonance imaging (fMRI) technology, attempted to capture traces of this subjective experience from the perspective of brain activity, quantifying the correlation between artistic abstraction and perceptual dispersion.
C. Durkin, M. Apicella, C. Baldassano, E. Kandel, & D. Shohamy, The Beholder’s Share: Bridging art and neuroscience to study individual differences in subjective experience, Proc. Natl. Acad. Sci. U.S.A. 122 (15) e2413871122, https://doi.org/10.1073/pnas.2413871122 (2025).
Durkin, Celia. The Beholder’s Share: Bridging Art and Neuroscience to Study Subjective Experience. Columbia University, 2023.
In one experiment, researchers attempted to quantify group differences induced by different types of art. The results showed that evaluations of abstract art varied widely among individuals; some argued that an abstract artwork should be exhibited tomorrow, while others believed the same artwork needed a year to qualify for a gallery. This was referred to in the paper as "spatiotemporal evaluation dispersion" and showed a significant positive correlation with participants' subjective evaluations of abstract works.
Figure 3: Behavioral experiment design, 840 online participants, each viewing one painting (21 paintings in total, 40 viewers per painting). Each participant imagined themselves as an art consultant; after seeing a painting, they judged whether it should be placed in "a gallery opening tomorrow" or "a gallery opening in one year," and whether it should be placed in a local museum or a museum in another state. Image source: PNAS
This groundbreaking research not only validated the behavioral manifestation of the "beholder's share" but also revealed, through neuroimaging techniques, behavioral evidence that abstract art indeed elicits more individualized cognitive processing. To ascertain what exactly happens behind this, researchers designed a two-stage experiment: the first stage involved behavioral studies to confirm the phenomenon, while the second stage used functional magnetic resonance imaging (fMRI) to capture real-time brain activity during thought processes.
In the experiment, researchers selected 21 paintings created by the same artist, categorized into representational art, indeterminate art, and abstract art. 30 ordinary individuals without art training were asked the same questions about these paintings: whether they thought the painting should be exhibited tomorrow or in a year, and they were asked to describe the painting in words. Participants had four seconds to think, while their brain activation areas were recorded by fMRI.
Figure 4: Neuro-response experiment design. Image source: PNAS
Neuroimaging data revealed a hierarchical processing mechanism of art perception. In the early stages of visual information processing, such as in the occipital pole and fusiform gyrus, various types of artworks activated similar regions, indicating that visual input itself did not cause differences. However, in higher-order cognitive processing stages, different types of art showed individualized differences. Abstract art elicited more inter-individual differences in neural activity, with the Default Mode Network (DMN) showing particularly prominent activity. The DMN is believed to be closely associated with autobiographical memory retrieval and scene simulation functions, meaning that abstract art is more likely to evoke individual experience, thus leading to a higher degree of subjective experience intervention in the "beholder's share." Even after controlling for overall activation levels, differences between abstract and representational art persisted, indicating that the differences were not due to some paintings causing stronger overall activation, but rather to different individual interpretation methods.
Figure 5: Relationship between abstract and representational art, and the degree of brain activity difference with abstract scores, as well as a scatter plot of abstractness scores and semantic differences. Image source: PNAS
Semantic analysis of text descriptions using natural language models showed that abstract artworks triggered significantly greater linguistic variation than representational works. This semantic diversity correlated positively with neural CSD in the precuneus and medial frontal cortex. Furthermore, the higher participants rated the "abstractness" of a painting, the greater the inter-individual variability in their brain responses and linguistic descriptions, indicating that abstractness is not merely an art classification label but a key variable influencing subjective experience.
In the emotional context regulation experiment, participants were asked to first recall a positive, negative, or neutral emotional memory, and then describe the artwork in words. The experiment found that participants who experienced positive emotional priming used more emotionally rich adjectives, while the negative emotional group preferred calm and restrained expressions. The interactive effect of abstract art and emotional priming was particularly prominent; with emotional induction, the semantic differences in descriptions of representational art were smaller. This reflects that abstract art provides a larger psychological space, allowing individuals to freely inject their emotional experiences and associations, therefore, after emotional induction, the subjective differences in abstract art were most significant.
Figure 6: A, Emotional states of each participant group. B. Correlation between emotional scores and recently experienced positive emotions. C. Correlation between emotional scores and recently experienced negative emotions. D. Schematic diagram of the emotional memory induction task. E. Percentage of positive emotional words in memory descriptions.
Differences in language style were also reflected in brain activity, especially in the default mode network regions. This indicates that subjective experience is not only influenced by the type of art but also regulated by the individual's current mental state. This finding supports that the default mode network is not only related to self-cognition but also participates in regulating subjective interpretations based on emotional memories.
Neuroscience Embraces Individuality,
No Longer Treats Subjectivity as Noise
Before discussing the significance of this research, let's first answer the opening question: why do Ghibli-style illustrations generated by ChatGPT still go viral years after the advent of AI drawing? From the perspective of emotional arousal, such works not only inherit the healing qualities of Ghibli animation, prompting viewers to recall positive emotional experiences, but also maintain concrete narrative features, capable of eliciting similar neural responses in most people, thus allowing them to break into the mainstream. If they were Gothic works, they would likely receive vastly different evaluations.
The Zen philosophy of "a flower, a world" perfectly echoes the "beholder's share" hypothesis across time and space. Traditional scientific research tends to eliminate individual differences, pursuing repeatable average effects, but this precisely overlooks the essence of subjective experience. If we view subjectivity as a meaningful source of information rather than noise to be removed, then we might better understand cognitive differences among individuals. "The beholder's share" is not just a concept of art appreciation; it is key to understanding the diversity and creativity of the human mind. Subjectivity itself is a meaningful cognitive signal, revealing how individuals use prior experience to construct their understanding of the world. The closely related default mode network may be intimately involved in cognitive processes such as creative thinking, memory reconstruction, emotional regulation, and future imagination.
Furthermore, if one wants to study subjective experience, abstract art combined with an open-ended textual description paradigm demonstrates unique advantages. Compared to traditional multiple-choice questions or rating scales, experimental designs that ask participants to freely describe their feelings about artworks, coupled with semantic vector analysis techniques in natural language processing, can effectively capture individual differences in linguistic representation. This provides a new path for studying individual differences. When such text data is cross-validated with synchronously collected neuroimaging data, researchers can trace the neurobiological origins of subjective experience differences, opening new dimensions for diversity research.
Art is not merely an object of beauty, but a bridge to the subjective mind. "The observer's share" reminds us that every viewer is a unique perceptual subject, and our task is to describe the world as it truly is using scientific methods. Building on this, future research may use analysis of patients' subjective descriptions of art to aid in diagnosing psychological disorders, for example, by predicting whether an adult subject suffers from depression based on neural activity while watching movie clips [1]. Additionally, art can be utilized to alleviate symptoms of mental illnesses and assist in the treatment of conditions such as PTSD or Alzheimer's disease.
In education, art can be used to stimulate students' subjective associations and promote creative thinking; generating text from images can also allow artists to successfully explore more ideas and assist them in creation. Harmoniously integrating human exploration with artificial intelligence utilization may lead to new creative workflows.
AI researchers can also build models to reproduce human subjective interpretation abilities for artworks. For example, using the same neural network (ResNet50) trained on different visual datasets to simulate the diversity of individual visual experiences in a population, and comparing the differences in cross-network activation between abstract and representational paintings [2]. This technical path echoes the widely applied AI style transfer, where a system transferring the brushstroke characteristics of Van Gogh's "Starry Night" to a photographic work is essentially performing a specific mode of top-down perceptual reconstruction.
Neuroscience research into art experience should be a win-win for both artists and neuroscientists. Both art and science attempt to understand and describe the world around us; they differ in their methods and communication. Art creates more memorable experiences by evoking emotions or understanding, while science provides tools for empirical observation and reasoning. Art is evolutionarily linked to the uniqueness of the human brain and to physiological impacts brought about by sexual selection.
Cognitive neuroscience research indicates that art is not merely a symbolic system but a multi-dimensional interactive process that integrates the creator's intent and the viewer's experience, with this interaction being jointly regulated by individual cognitive background and neural plasticity. The revelation of the specificity of artistic experience opens a window for us to understand the complexity of human experience, allowing us to glimpse what makes our minds unique. This is particularly insightful in the age of big data; while algorithms attempt to simplify human nature into data vectors, art reminds us of the irreproducible richness of the mind.
Relationship between art and neuroscience. Image source: https://pmc.ncbi.nlm.nih.gov/articles/PMC2815940/
It is precisely because artistic experience is subjectively constructed that it is indispensable for personal physical and mental well-being. Curiosity, surprise, awe—all these common traits in art creators or observers—are very important for a person with a rich inner life. The dynamic network composed of approximately 86 billion neurons in our brain, with its plasticity, determines how the quality of environmental stimuli shapes cognitive function. The richer the environment, the stronger the sensory experience, in a way that feels safe and often novel, facilitating our brain's restructuring to support high-speed operation in the future, much like a computer that has been on for too long needs disk defragmentation. The appreciation and creation of art should become part of healthy lifestyle habits.
[1]https://academic.oup.com/pnasnexus/article/3/3/pgae052/7618478?login=false
[2]https://jov.arvojournals.org/article.aspx?articleid=2792703
[3]https://aeon.co/videos/on-the-beholders-share-how-past-experience-influences-our-perception-of-art