AESTHETICS OF ARTIFICIAL INTELLIGENCE ART

The study „Aesthetics of Artificial Intelligence Art“ explores the connection between AI art and computational neuroaesthetics. It adopts three positions: theoretical, analytical, and experimental. The theoretical point of view is grounded in the discourse of art and AI aesthetics, the analytical one focuses on studying contemporary artistic practices, and the experimental is a practical-speculative approach to exploring art and aesthetic possibilities. The research aims to discern if non-human AI art can be created and under what conditions. An anticipated significant contribution of this study is the accelerated application of computational neuroaesthetics, influencing methodological approaches beyond this research project.

Updated project proposal: May 2023

AESTHETICS OF ARTIFICIAL INTELLIGENCE ART

Tomáš Marušiak MFA

Masaryk University Brno, Faculty of Arts, Department of Musicology (Czech Republic)

Cultcode-Institute of Visual Art (Slovak Republic )

ABSTRACT

The research project „Aesthetics of Artificial Intelligence Art“ examines the relationships between artificial intelligence art and computational neuroaesthetics. Research approaches are articulated from three basic positions. First, the theoretical position is situated in the discourse of art and aesthetics of artificial intelligence. The analytical position relies on the study of contemporary artistic practice and will identify artistic strategies in the subject area, mainly about the objectives of the project. The experimental position represents a practical-speculative approach to conducting research into the possibilities of art and aesthetics, which can also be understood as artistic research. The research aims to provide an answer to the question of whether it will be possible to create non-human artificial intelligence art and under what conditions. To achieve research goals, but also as a significant contribution, the acceleration of application possibilities of computational neuroaesthetics can be considered, the benefits of which have the potential to influence related methodological approaches even beyond the scope of this research project.

THEORETICAL BASIS AND CURRENT STATE OF SCIENTIFIC KNOWLEDGE

The description of theoretical starting points and the current state of scientific knowledge is important to divide based on the three positions already mentioned. Each of the positions is connected with the others but requires a different description of the state of research and methodological approach. The first position is situated in the discourse of art and aesthetics of artificial intelligence, neurosciences, and computer science in order to cover the cores of the mentioned scientific disciplines and their intersections as comprehensively as possible. The second position is based on the study of art practice in relation to artificial intelligence art and identification of artistic strategies, which are a significant source of inspiration when conducting experiments and studies. The third position represents practical-speculative approaches and is based on theoretical starting points formulated based on research in the first two positions.

2.1. THE BASIS OF THEORETICAL POSITION

The theoretical position aims to analyze the issue within a broader framework delimited by scientific as well as artistic research, as well as by formulating interdisciplinary strategies that reach into various areas of knowledge. For the purpose of verifying epistemic attitudes or positions offered in the AI Art World model, it is appropriate not to shy away from creating experimental procedures. The AI Art World model for the purposes of the project represents a paraphrase of Danto´s definition of Art World[1], in which AI art dominates. I place significant emphasis on the transposition of artistic research[2], even if appropriated into the environment of scientific procedures, with a clearly defined methodology based on exact and reliably readable evidence. On the other hand, it is important not to succumb to dataistic fanaticism[3] and the idea of a world consisting of data form. Reflections on the presented topics as well as on the outputs from research have the common potential to bring a new perspective on interdisciplinary dialogue in this space at multiple levels and to achieve research goals as well as expect the emergence of new impulses.

Figure 1. Reflection of the theoretical position research discourse framework.

Source: Archive of Tomaš Marušiak and AFP/Getty Images.

2.2. ARTIFICIAL INTELLIGENCE ART

The art, and therefore also the aesthetics of artificial intelligence art, began to be formulated at the beginning of the second decade of the 21st century in visual art, associated with the most prominent authors, such as Mario Klingemann, Memo Akten, Mike Tyka, Refik Anadol, Gene Kogan, Lauren Lee McCarthy, Joy Buolamwini, Stephanie Dinkins, Jake Elwes, Trevor Paglen. However, we must look for the beginnings of the idea of AI art as an artistic construction among the pioneers[4] of generative art. Among the theorists who have significantly influenced the discourse are primarily Lev Manovich, Joanna Zylinska, in connection with publications AI Art[5] or The Future media[6]. Matias Del Campo made a significant contribution to the theory of AI architecture with publications Machine Hallucinations: Architecture and Artificial Intelligence: Architectural Design[7] and Neural Architecture: Design and Artificial Intelligence[8]. In addition to all organizations supporting the emerging AI art, it is important to point out initiatives such as Ars Electronica, European ARTificial Intelligence Lab and the publication The Practice of Art and AI[9].

2.3. AESTHETICS OF ARTIFICIAL INTELLIGENCE ART: ATTEMPTS AT DEFINITION

I consider it important to describe Manovich’s reflection from 2019, Defining AI Arts: Three Proposals[10], on the current state of artificial intelligence and visual art. It is undeniable that we find ourselves in a situation that relies on artificial intelligence and that simultaneously fulfills the desires of what we as humans cannot achieve. The first condition he puts forward for using AI is to teach it to understand the history of art. In general, however, it can be said that we know how to teach a computer to understand social relationships and art history, or better said, nothing prevents us from starting to recognize it. In relation to the cultural transformation under the influence of emerging utilitarian and entertaining AI, as well as evolving trends in the creation of Foundation Models[11] (CRFM), I see the possibility of creating a history and theory of art by using the mentioned models as already feasible.

Manovich’s second attempt admits the existence of at least three points when the human author makes explicit decisions and directs what the computer should do. Manovich concludes that what defines AI is the type of control over the processes performed by man. The third attempt, therefore, is based on the idea of a computer that can learn the structure of the world differently than when we force computers to create like us today. As a starting point, the concept is offered to teach the computer what we humans cannot do. This means primarily to teach AI to create something that surprises us with new aesthetics, and at the same time teaches us to perceive it.

Manovich’s third attempt, although it lies in the realm of speculation, is at the same time the creation of a starting track for opening the discourse on artificial aesthetics[12]. According to Manovich and Arielli, it is important to understand the aesthetics of AI as deepening creative processes, as well as understanding and perceiving cultural artifacts. I believe that expanding the scope of aesthetics, which includes natural and man-made objects and experiences, as well as experiences created by a machine, is a fundamental starting point. Therefore, such a definition also affects the special area of „beauty“, such as the beauty of mathematics or the beauty of AI itself. According to Manovich, the encounter of AI with aesthetics is key as a harbinger of a change in perspective on aesthetics, which has so far been only a human domain.

In general, it can be established, even without special proof, that if aesthetics is already a non-human domain, it must be in interaction with a similarly complex and at least partially artificial system, as we know it in connection with human aesthetics. However, we still have to look at such interaction through a speculative model, as a certain kind of general artificial intelligence. I note that for AI generating an image or music composition, we are still moving in the environment of weak or narrow AI. Based on this, it is important to involve computational neuroaesthetics in the modeling processes, which can be understood as an effort to understand the neuronal basis of aesthetic experiences, using mathematical models and computational algorithms to analyze and understand the mechanisms that underlie aesthetic experiences and human experience.

2.4. CONSCIOUSNESS: THE MODEL OF ARTIFICIAL CONSCIOUSNESS AS A STARTING POINT

I see the articulation of the predictive Model of the AI Art World in the complexity in which several parallel systems of natural and artificial consciousness interact. In addition to the systems of human natural learning AI[13], I consider at least 3 views on the concept of consciousness to be key lately. The first represents the theoretical framework of Karl Friston and parallels it with predictive encoding in AI[14]. The second view is through the theory of integrated information and the third parallel with the conceptualization of the theoretical framework of consciousness by Feinberg and Mallat[15], which assumes that the perception of an image, which is necessary for the existence of primary consciousness, is not only in mammals but also all vertebrates and fish.

Predictive coding according to Karl Friston[16] is a theoretical framework in neuroscience that is based on the brain constantly making predictions about its environment and updating its predictions based on incoming sensory information. Predictive coding[17] in artificial intelligence (AI) and machine learning helps algorithms learn from data and make predictions. In predictive coding, an initial prediction is made and then the actual result is compared with the prediction. Any differences between the prediction and the actual result are used to adjust the parameters of the model, which in turn improves the accuracy of future predictions.

The theory of integrated information[18] (IIT) is a theoretical framework developed by Giulio Tononi, which seeks to explain the essence of consciousness. According to IIT, consciousness arises as a result of integrated information generated by the brain. Information is considered integrated if it cannot be broken down into separate parts and therefore must or can be readable when examining the brain. IIT also suggests that consciousness is a property of any system that generates integrated information. This means that it is not limited to biological systems, but may also exist in artificial systems that are capable of generating integrated information.

Figure 2. It represents the positions of various theoretical starting points about both natural and artificial consciousness at the level of the speculative field of potential connections. A, Friston’s and Kiebel’s principle of free energy as the basis for the theory of predictive coding. B, The theory of integrated information. C, Giulio Tononi’s scientific view suggests that consciousness might be exclusively a human domain. D, The AI principle of predictive coding. E, Tme feasibility of experiments about the research project.

Source: Archive of Tomáš Marušiak and AFP/Getty Images.

2.5. COMPUTATIONAL NEUROAESTHETICS

A brief history of neuroaesthetics can be described by its most significant conceptual models. As the first, I introduce Zeki’s model and Ramachandran and Hirstein’s model. Semir Zeki[19] focused attention on visual art and its physical properties that are reflected in our brains. In the 20th century, Zeki explored the ideas of philosophers, neurologists, and artists. Based on their views, he suggested that the function of art is an extension of the visual brain. Zeki promotes the idea that artists unconsciously use techniques to create visual art to study the brain[20]. Ramachandran[21] and Hirstein combined an evolutionary approach with neurophysiological evidence and proposed a model to explain the aesthetic experience of visual art[22].

Next, I mention Leder’s visual aesthetic model based on previous findings in psychology and neuroscience [23]. This model is based on processing information about aesthetic appreciation, which is divided into five stages: context and model input, perceptual analysis, implicit memory integration, explicit classification, cognitive mastering, and evaluation. In 2003, Chatterjee proposed a linear processing model based on visual and aesthetic experience[24]. The principle is built on the idea that the aesthetic experience emerges from the interaction of three neural systems: the sensorimotor system, the emotional-evaluative system, and the system of knowledge meaning. This model is useful for linking different aspects of the cognitive processing stages of the aesthetic experience to specific brain structures. In his 2011 article, „Neuroaesthetics: A Coming of Age Story“[25], Chatterjee indirectly defines the research framework of computational neuroaesthetics.

I consider the 2020 article „Review of computational neuroaesthetics“[26] to be significant, in which the authors highlight the need for interdisciplinarity between aesthetics, neuroscience, and computer science. Formulations of starting points appealing, for example, to the need for machine learning involvement in some way heralded the imminent advent of AI processes that can more reliably read human cognitive processes. On May 1, 2023, Nature published the article „Semantic reconstruction of continuous language from non-invasive brain recordings“[27], which refers to the feasibility of possibly reading human thoughts with the help of Foundation Models and functional magnetic resonance. A similar principle is included in the proposed methodologies but using a portable neuroimaging device with the support of Socratic models.

Figure 3. It represents the definitional position of computational neuroaesthetics, which intersects with the proposal of the definition of Artificial Aesthetics by Manovich and Arielli.

Source: Archive of Tomáš Marušiak

2.6. ATTEMPT TO DEFINE: THE AESTHETICS OF ARTIFICIAL INTELLIGENCE ART

The ambition of the project is to create a research space at the intersections of neuroscience, aesthetics, artificial intelligence, computer science, and artistic research. By subsuming these theoretical premises in light of the current state of research, I propose a model defining the Aesthetics of Art of Artificial Intelligence for the purpose of constructing a research project. The model is derived from the auxiliary definition of computational neuroaesthetics, which overlaps with the conceptual framework of Artificial Aesthetics by Manovich and Arielli.

The Aesthetics of Artificial Intelligence Art (AAI), is an area of computational aesthetics aimed at understanding and creating artificial systems capable of analyzing and producing aesthetic experiences and facilitating their transfer between humans and machines. This multidisciplinary field combines computer science, artistic practice, art science, neuroscience, psychology, and philosophy. AI technical means are used in AAIA to analyze, understand, and simulate human thinking, perception of simulation processes of consciousness, as well as the creation of models of the World of Art of Artificial Intelligence. Research in the field of AAIA promotes the development of intelligent systems that can enhance creative activity, but also allows for a deeper understanding of the processes of natural and artificial consciousness.

FORMULATION OF THE RESEARCH PROBLEM

The fundamental platform for formulating the central research problem is primarily to detect significant factors, in relation to the theoretical foundations already mentioned, that can create a model of artificial consciousness capable of creating art as a unique identity depending on their mutual interaction. The research problem can be established at several levels. At the first level, it is the description of the problem. Is the AIArtWorld concept feasible, including artificial forms of actors in the AI social field of art such as author-producer-recipient? The second relational-causal level represents the relationship between humans and the AIArtWorld. What is the role of humans in the AIArtWorld concept, how will it be influenced by humans, if at all? The third level is already formulated as a problem of shaping speculative models of future AI Art aesthetics.

It is also necessary to point out the risks associated with the relatively strong and rapid onset of progress in the field of AI. The rapid trends mentioned will likely change the perspective during the implementation of the entire research, especially in positions directly related to specific technologies and social moods.

RESEARCH GOALS

The main goal of the research is to provide an answer to the question of whether it will be possible to create non-human art of artificial intelligence and under what conditions. Equally important in the short term is to set goals that will establish the methodology among several scientific disciplines and artistic research. To achieve the research goals, but also as a significant contribution, it is possible to consider the important acceleration of application possibilities of computational neuroaesthetics, the benefits of which have the potential to influence related methodological approaches even beyond the framework of this research project. Therefore, it can be assumed that unpredictable goal proposals will emerge within the outlined conceptual framework during the implementation of the research, considering the dynamics of AI development.

THEORETICAL POSITION

The research methodology about the theoretical position is designed in close connection with the design of the university course „AI Art Strategy“. The research will be conducted with a focus on 5 basic areas. As the first area, I mention the research of AI Art in the current context of culture, science, and technology. The following is an area oriented toward the roots of AI Art about conceptual art and post-humanism. There’s an area of artistic strategies about application settings, technological trends, and artistic strategies that carry out a certain form of artistic research. The quintet is completed with an area related to design and architecture.

CONTEMPORARY ART PRACTICE STUDIES

The study of contemporary artistic practice is mainly focused on the analysis and subsequent identification of artistic strategies and projects that will serve to derive experiments. This study can also be understood as an intersection between theoretical and experimental positions. The development of artificial consciousness about AI is highlighted as one of the research problems. In this context, I select works for the derivation of future experiments that have a strong infra-experimental element in the form of artistic research and at the same time operate with consciousness or its speculative model transferable to AI. For example, I mention: The Changing Room[28] from 2021 by Lauren McCarthy. It is an AI-controlled installation that influences your feelings. Participants choose one of more than 200 emotions. The algorithm responds and induces the chosen feeling simultaneously in everyone in the room. The installation is a software platform that can be reconfigured to any space and network. Someone[29] from 2019 by Lauren McCarthy is a project representing a human version of Amazon Alexa. For two months in 2019, custom-designed smart devices, including cameras, microphones, lights, and other devices, were installed in the homes of four participants across the United States. The 205 Hudson Gallery in NYC housed a command center where visitors could peek into the four households via laptops, watch them, and remotely control their network devices.

Figure 4. Principle of selection-analysis of artistic strategies for the derivation of future experiments.

Source: Archive of Tomáš Marušiak

4.3. EXPERIMENTAL APPROACH TOWARDS NEUROAESTHETICS

This section is dedicated to technical imaging means and their significant contribution to the project. I will try to introduce 4 basic methods that I will use in the implementation of experiments. The mentioned methods are chosen based on minimal invasiveness, and mobility and also considering the external data analysis in research centers outside of my university lab that deals with neuroaesthetics and is unfeasible within the possibilities.

5.3.1. FUNCTIONAL NEAR-INFRARED SPECTROSCOPY (FNIRS)

fNIRS is a non-invasive neuroimaging technology that uses light in the near-infrared region to measure changes in the concentration of oxygenated and deoxygenated hemoglobin in the brain. These changes indicate neuronal activity and can be used to study brain function in response to various stimuli or tasks. fNIRS is often used as an alternative to other neuroimaging techniques, such as fMRI because it is portable and does not require participants to lie still inside a scanner. fNIRS can be used in a wide range of applications including cognitive neuroscience, clinical research, and human-machine interactions.

5.3.2. FACIAL EXPRESSION RECOGNITION (FER)

I consider emotion recognition from facial expressions to be the most comfortable about the recipient and usable as both an analytical and intervention method. Facial expressions are parts of communication where the linguistic structure of pronounced silences of syllables is summarized. I believe that the data obtained using this method are applicable for training large language models. Face recognition [30] is often an emotional experience for the brain, and the amygdala as well as the fusiform gyrus are highly involved in the recognition process.

5.3.3. SYNTHESIS OF SOCRATIC MODELS AND NEUROIMAGING METHODS

Socratic models [31], hereafter SM, provide a framework in which it is possible to compile several large pretrained models without the need for training to perform new subsequent multimodal tasks, using language – through challenge. Particularly important for use in the proposed research is that SM is useful for improving the accuracy and reliability of machine learning models in areas where there is a lot of ambiguity or uncertainty, such as natural language processing and image recognition. There is the possibility of learning with a human teacher who interacts with the model and asks it questions about the data it is evaluating. By asking questions, the teacher can lead the model to a more accurate understanding of the data and help it avoid common mistakes and biases. Based on this, I propose a speculative model (Figure 5.) that implements fNIRS, FER or electroencephalography methods for more reliable reading of emotional or other states caused by the visceral activity of the examined subject.

Figure 5. Example of perceptual tasks as a speculative case study of Socratic models with visual language models (e.g., video recordings), large language models (e.g., GPT-3), audio language models (e.g., Wav2CLIP, Speech2Text), and neuroimaging methods (e.g., EEG, fNIRS, EEG, etc.) is presented. The principle of video retrieval, generating open-ended responses to context-based reasoning questions with an evaluation of neurobiological processes (e.g., estimation of experienced emotions, etc.), and predicting future activities is shown. Socratic models can provide meaningful results for complex tasks in computationally demanding domains of computer vision without the need for fine-tuning the model architecture.

Source: Archive of Tomaš Marušiak and Timothy A. Clary—AFP/Getty Images

5.3.4. IMPLEMENTATIONS

Implementations of the experiments, as outlined in this section, represent one of the fundamental research tasks. Given this importance, it is essential to establish an implementation framework from the outset, which will cover the theoretical preparation of pilot studies and their actual implementation. I anticipate a series of project proposals that will be feasible to conduct at HUMELab and open projects that will be freely available for implementation at other research centers.

5.3.4.1. HUMELAB

The majority of experimental research studies are planned to be conducted at HUMELab, a unique facility of Masaryk University’s Faculty of Arts‘ innovative research infrastructure, equipped with top-notch technology for experimental research, including the neuroimaging technologies mentioned earlier. Since November 2022, a significant part of my efforts has been directed toward mapping HUMELab’s capabilities and their implementation in research projects. The first proposed project is a study called AI Self Style Empathy[32], which examines the cortical brain activity of a visual artist recorded through fNIRS in response to stimuli – works of art created through AI learning from his visual style. This study was derived from the already implemented study Does aesthetic judgment on face attractiveness affect neural correlates of empathy for pain?[33]. It is hypothesized that if the attractiveness of the generated artwork is high in the author’s subjective evaluation, then the artist will accept as his own a work that he did not create but is only inspired by it. Another three proposals are in the conceptualization phase.

5.3.4.2. OPEN PROJECTS

The term open projects covers all experiment concepts that cannot be implemented through one’s own or mediated research infrastructure due to time constraints, material or other technical obstacles. At this stage, it is necessary to account for risks in feasibility, and that’s why I propose to proceed with the publication of research projects with the option for open implementation by other research teams. The conditions for implementation by third parties, as well as the legal regulation of the use of the experiment concept as a work to which special protection applies, will form part of the publication of experiment concepts.

5.3.4.3. EXPERIMENTAL POSITION TOWARDS SPECULATIVE MODELS

The estimated feasibility of approximating a model of non-human art or AI Art World encounters a conceptual problem, which is, for instance, the articulation of conditions for interaction relationships between humans and AI Art World systems. It is also necessary to highlight the overlap between natural and artificial forms of actors in roles such as author-producer-recipient. In the current state of development, I have decided to use a method of predictive market analysis. As an example, I mention the Metaculus predictive market, which is used to gather predictions about the future. Users can predict a wide range of topics including science, technology, politics, and culture, and view aggregate community predictions. I anticipate that based on these analyses, it will be possible to create a predictive model of the AI Art World.

References

[1] DANTO, Arthur. The Artworld. The Journal of Philosophy [online]. 1964, 61(19) [cit. 2023-05-11]. ISSN 0022362X. Dostupné z: doi:10.2307/2022937

[2] SCHWAB, Michael. “Introduction.” Transpositions: Aesthetico-Epistemic Operators in Artistic Research, edited by Michael Schwab, Leuven University Press, 2018,

[3] Yuval Harari warns humans will be „hacked“ if artificial intelligence is not globally regulated. Cbsnews.com [online]. 2021. Dostupné z: https://www.cbsnews.com/news/yuval-harari-sapiens-60-minutes-2021-10-29/

[4] MILLER, Arthur I. The artist in the machine: the world of AI-powered creativity / Arthur I. Miller. 2019. ISBN 9780262539623.

[5] ZYLINSKA, Joanna, 2020. AI art : machine visions and warped dreams. London: Open Humanities Press. ISBN 9781785420856.

[6] ZYLINSKA, Joanna. The Future of Media. 1. Goldsmiths Press, 2022. ISBN 978-1-913380-14-4.

[7] DEL CAMPO, Matias, ed. Machine Hallucinations: Architecture and Artificial Intelligence: Architectural Design. 1. Hoboken, New Jersey, USA: Wiley, 2022. ISBN 978-1119748847.

[8] DEL CAMPO, Matias, ed. Neural Architecture: Design and Artificial Intelligence. 1. USA: Applied Research & Design, 2022. ISBN 978-1951541682.

[9] GALIROW, Jessica, SCHMIEDERER, Jutta, ed. The Practice of Art and AI. 1. Linz: Ars Electronica,European ARTificial Intelligence Lab, 2021. ISBN 9783775751421.

[10] MANOVICH, Lev. Defining AI Arts: Three Proposals. Manovich.net [online]. 2019 [cit. 2021-03-03]. Dostupné z: http://manovich.net/index.php/projects/defining-ai-arts-three-proposals

[11] Introducing the Center for Research on Foundation Models (CRFM) [online]. Human-Centered AI (HAI),Stanford University, 2023 [cit. 2023-04-23]. Dostupné z: https://hai.stanford.edu/news/introducing-center-research-foundation-models-crfm

[12] MANOVICH, Lev a Arielli EMANUALE. Artificial Aesthetics A Critical Guide to AI, Media and Design: A Critical Guide to AI, Media and Design [online]. 1. 2021 [cit. 2022-03-08]. Dostupné z: https://www.academia.edu/

[13] LAKE, Brenden M., Tomer D. ULLMAN, Joshua B. TENENBAUM a Samuel J. GERSHMAN. Building machines that learn and think like people. Behavioral and Brain Sciences [online]. 2017, 40 [cit. 2023-04-23]. ISSN 0140-525X. Dostupné z: doi:10.1017/S0140525X16001837

[14] MILLIDGE, Beren, Alexander TSCHANTZ a Christopher L. BUCKLEY. Predictive Coding Approximates Backprop Along Arbitrary Computation Graphs. Neural Computation [online]. 2022, 34(6), 1329-1368 [cit. 2022-11-20]. ISSN 0899-7667. Dostupné z: doi:10.1162/neco_a_01497

[15] JON M. MALLATT. The Ancient Origins of Consciousness: How the Brain Created Experience. 2016. ISBN 9780262034333.

[16] FRISTON, Karl a Stefan KIEBEL. Predictive coding under the free-energy principle. Philosophical Transactions of the Royal Society B: Biological Sciences [online]. 2009, 364(1521), 1211-1221 [cit. 2022-11-20]. ISSN 0962-8436. Dostupné z: doi:10.1098/rstb.2008.0300

[17] MILLIDGE, Beren, Alexander TSCHANTZ a Christopher L. BUCKLEY. Predictive Coding Approximates Backprop Along Arbitrary Computation Graphs. Neural Computation [online]. 2022, 34(6), 1329-1368 [cit. 2022-11-20]. ISSN 0899-7667. Dostupné z: doi:10.1162/neco_a_01497

[18] TONONI, Giulio a Christof KOCH. Consciousness: here, there and everywhere?. Philosophical Transactions of the Royal Society B: Biological Sciences [online]. 2015, 370(1668) [cit. 2022-11-20]. ISSN 0962-8436. Dostupné z: doi:10.1098/rstb.2014.0167

[19] ZEKI, Semir. Inner Vision: An Exploration of Art and the Brain. 1. Oxford University Press, 2000. ISBN 978-0198505198.

[20] ZEKI, Semir. Splendors and Miseries of the Brain: Love, Creativity, and the Quest for Human Happiness. 1. London: Wiley-Blackwell, 2008. ISBN 978-1405185578.

[21] NADAL, Marcos a Martin SKOV. Neuroesthetics. In: International Encyclopedia of the Social & Behavioral Sciences [online]. Elsevier, 2015, 2015, s. 656-663. ISBN 9780080970875. Dostupné z: doi:10.1016/B978-0-08-097086-8.56027-9

[22] RAMACHANDRAN, V.S. a William HIRSTEIN. The science of art: a neurological theory of aesthetic experience. Journal of Consciousness Studies. Imprint Academic, 1999, 1999(6-7), 15-51(37). ISSN 1355-8250.

[23] LEDER, Helmut a Marcos NADAL. Ten years of a model of aesthetic appreciation and aesthetic judgments: The aesthetic episode – Developments and challenges in empirical aesthetics. British Journal of Psychology [online]. 2014, 105(4), 443-464. ISSN 00071269. Dostupné z: doi:10.1111/bjop.12084

[24] CHATTERJEE, Anjan a Oshin VARTANIAN. Neuroaesthetics. Trends in Cognitive Sciences [online]. 2014, 18(7), 370-375 . ISSN 13646613. Dostupné z: doi:10.1016/j.tics.2014.03.003

[25] CHATTERJEE, Anjan. Neuroaesthetics: A Coming of Age Story. Journal of Cognitive Neuroscience [online]. 2011, 23(1), 53-62. ISSN 0898-929X. Dostupné z: doi:10.1162/jocn.2010.21457

[26] LI, Rui a Junsong ZHANG. Review of computational neuroaesthetics: bridging the gap between neuroaesthetics and computer science. Brain Informatics [online]. 2020, 7(1). ISSN 2198-4018. Dostupné z: doi:10.1186/s40708-020-00118-w

[27] TANG, Jerry, Amanda LEBEL, Shailee JAIN a Alexander G. HUTH. Semantic reconstruction of continuous language from non-invasive brain recordings. Nature Neuroscience [online]. 2023, 26(5), 858-866. ISSN 1097-6256. Dostupné z: doi:10.1038/s41593-023-01304-9

[28] MCCARTHY, Lauren. Lauren-mccarthy: SOMEONE [online]. 2022. Dostupné z: https://lauren-mccarthy.com/ The-Changing-Room

[29] MCCARTHY, Lauren. Lauren-mccarthy: SOMEONE [online]. 2022. Dostupné z: https://lauren-mccarthy.com/SOMEONE

[30] ROBSON, Matthew, Romina PALERMO, Linda JEFFERY a Markus NEUMANN. Ensemble Coding of Face Identity in Congenital Prosopagnosia. Journal of Vision [online]. 2017, 17(10) [cit. 2022-03-09]. ISSN 1534-7362. Dostupné z: doi:10.1167/17.10.624

[31] CHANG, Edward Y. Prompting Large Language Models With the Socratic Method. In: 2023 IEEE 13th Annual Computing and Communication Workshop and Conference (CCWC) [online]. IEEE, 2023, 2023-3-8, s. 0351-0360 [cit. 2023-05-06]. ISBN 979-8-3503-3286-5. Dostupné z: doi:10.1109/CCWC57344.2023.10099179

[32] MARUŠIAK, Tomáš. AI Self Style Empathy. MUNI-Brno, 2022.

[33] BALCONI, M., N. KOPIS a L. ANGIOLETTI. Does aesthetic judgment on face attractiveness affect neural correlates of empathy for pain? A fNIRS study. Experimental brain research [online]. 2020, 238(9), 2067-2076 [cit. 2022-11-30]. ISSN 14321106. Dostupné z: doi:10.1007/s00221-020-05867-y