Is my "red" your "red"?: Unsupervised alignment of qualia structures via optimal transport

Kawakita, Genji; Zeleznikow-Johnston, Ariel M; Tsuchiya, Naotsugu; Oizumi, Masafumi

doi:10.31234/osf.io/h3pqm

Cited by 17 publications

(20 citation statements)

References 35 publications

(53 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Recent breakthroughs in this domain demonstrate that a massive number of similarity relationships among words in one language (e.g., English) can find quite accurate correspondence with those in another language (e.g., Japanese) without knowing any labels (Alvarez-Melis and Jaakkola, 2018 ). Such a method could be introduced in consciousness research, helping find equivalence, and adjoint relationships between structures of qualia and behavioral reports (Kawakita et al, 2023 ).…”

Section: Discussionmentioning

confidence: 99%

An adjunction hypothesis between qualia and reports

2023

Self Cite

View full text Add to dashboard Cite

What are the nature of the relationship among qualia, contents of consciousness, and behavioral reports? Traditionally, this type of question has been only addressed via a qualitative and philosophical approach. Some theorists emphasize an incomplete and inaccurate nature of reports of one's own qualia to discourage formal research programs on qualia. Other empirical researchers, however, have made substantial progress in understanding the structure of qualia from such limited reports. What is the precise relationship between the two? To answer this question, we introduce the concept of “adjoint” or “adjunction” from the category theory in mathematics. We claim that the adjunction captures some aspects of the nuanced relationships between qualia and reports. The concept of adjunction allows us to clarify the conceptual issues with a precise mathematical formulation. In particular, adjunction establishes coherence between two categories that cannot be considered equivalent, yet has an important relationship. This rises in empirical experimental situations between qualia and reports. More importantly, an idea of adjunction naturally leads to various proposals of new empirical experiments to test the predictions about the nature of their relationship as well as other issues in consciousness research.

show abstract

Section: Discussionmentioning

confidence: 99%

An adjunction hypothesis between qualia and reports

2023

Self Cite

View full text Add to dashboard Cite

show abstract

“…This parallelism is not limited to concepts and can also be observed on the sensory knowledge that can be captured from language such as colours. For example, Kawakita et al (2023) showed a striking similarity between human ratings between colours and the corresponding similarity ratings produced by GPT-4.…”

Section: Large Language Models (Llms)mentioning

confidence: 94%

Language models and psychological sciences

Sartori,

Orrù

2023

Front. Psychol.

View full text Add to dashboard Cite

Large language models (LLMs) are demonstrating impressive performance on many reasoning and problem-solving tasks from cognitive psychology. When tested, their accuracy is often on par with average neurotypical adults, challenging long-standing critiques of associative models. Here we analyse recent findings at the intersection of LLMs and cognitive science. Here we discuss how modern LLMs resurrect associationist principles, with abilities like long-distance associations enabling complex reasoning. While limitations remain in areas like causal cognition and planning, phenomena like emergence suggest room for growth. Providing examples and increasing the dimensions of the network are methods that further improve LLM abilities, mirroring facilitation effects in human cognition. Analysis of LLMs errors provides insight into human cognitive biases. Overall, we argue LLMs represent a promising development for cognitive modelling, enabling new explorations of the mechanisms underlying intelligence and reasoning from an associationist point of view. Carefully evaluating LLMs with the tools of cognitive psychology will further understand the building blocks of the human mind.

show abstract

“…All examples shown in this paper can be tested in the tutorial notebooks to help new users get started. In addition, another example using the similarity judgment data of 93 colors previously reported in [6] is also available. The processed data used to generate the results presented in this study are also available in the "data" folder of the same GitHub repository.…”

Section: Availability and Future Directionsmentioning

confidence: 99%

“…For example, we can consider the alignment between neural activity and behavior or a model. This application is important for assessing how well a given neural network model captures neural responses in the brain or how well it explains behavior in humans or other animals [7,[24][25][26][27][28][29]. Moreover, we can also consider the alignment between behavior and neural activity, which is important for identifying neural correlates that explain a given behavior in humans or other animals [14,[30][31][32][33][34][35].…”

Section: Availability and Future Directionsmentioning

confidence: 99%

Unsupervised Alignment in Neuroscience: Introducing a Toolbox for Gromov-Wasserstein Optimal Transport

Sasaki,

Takeda,

Abe

et al. 2023

Preprint

Self Cite

View full text Add to dashboard Cite

Gromov-Wasserstein optimal transport (GWOT) has emerged as a versatile method for unsupervised alignment in various research areas, including neuroscience, drawing upon the strengths of optimal transport theory. However, the use of GWOT in various applications has been hindered by the difficulty of finding good optima, a significant challenge stemming from GWOT's nature as a non-convex optimization method. It is often difficult to avoid suboptimal local optima because of the need for systematic hyperparameter tuning. To overcome these obstacles, this paper presents a user-friendly GWOT hyperparameter tuning toolbox (GWTune) specifically designed to streamline the use of GWOT in neuroscience and other fields. The toolbox incorporates Optuna, an advanced hyperparameter tuning tool that uses Bayesian sampling to increase the chances of finding favorable local optima. To demonstrate the utility of our toolbox, we first illustrate the qualitative difference between the conventional supervised alignment method and our unsupervised alignment method using synthetic data. Then, we demonstrate the applicability of our toolbox using some typical examples in neuroscience. Specifically, we applied GWOT to the similarity structures of natural objects or natural scenes obtained from three data domains: behavioral data, neural data, and neural network models. This toolbox is an accessible and robust solution for practical applications in neuroscience and beyond, making the powerful GWOT methodology more accessible to a wider range of users. The open source code for the toolbox is available on GitHub. https://oizumi-lab.github.io/GWTune/

show abstract

Is my "red" your "red"?: Unsupervised alignment of qualia structures via optimal transport

Cited by 17 publications

References 35 publications

An adjunction hypothesis between qualia and reports

An adjunction hypothesis between qualia and reports

Language models and psychological sciences

Unsupervised Alignment in Neuroscience: Introducing a Toolbox for Gromov-Wasserstein Optimal Transport

Contact Info

Product

Resources

About