Quantifying Qualia – Aesthetic Machine Attention in Resisting the Objectifying Tendency of Thought

Abstract

My interdisciplinary doctoral thesis Quantifying Qualia – Aesthetic Machine Attention in Resisting the Objectifying Tendency of Thought, conducted at the Department of Art and Media at Aalto University, explores human and algorithmic perception. While language-based approaches are widely developed and utilized in machine learning today, the thesis explores the ethical potential of alternative modes of perception to be manifested in machines and proposes the concept of aesthetic attention to invite perceptual variations from phenomena through how they resonate across the senses. Psychologist Daniel Stern suggests that this dynamic nature of experience, arising from embodiment, represents the earliest stage of development. Consequently, it serves as the primary means for interpersonal communication and also expressing inner experiences later in life. Additionally, affective and aesthetic expressions can be viewed as being rooted in these vitality forms described by Stern.

The thesis argues that aesthetically oriented attention has the potential to reorganize perception by delaying the categorical determination of an experience. At the core of my research is the idea that the narrowed cognitive repertoire resulting from perceptual biases can be altered with perceptual strategies aiming to broaden the receptivity for sensory knowledge.

My thesis consists of three peer-reviewed articles published in interdisciplinary edited volumes and journals, along with one peer-reviewed unpublished article. These articles redefine philosophical concepts such as aesthetic attention and qualia, making them computable. As a result, a method was developed in interdisciplinary collaboration to generate asemic stimuli algorithmically. This approach also led to the establishment of a research platform that seamlessly integrated both artistic and quantitative research.

The artistic conclusion of my thesis is a research process utilizing the platform. During this process, asemic stimuli were annotated with artistic expressions as opposed to the traditional method of using verbal categories for annotating. Multimodal expressions established aesthetic data for a machine attention model to perceive beyond categories.

With the research process, I demonstrated how the development of machine learning models that incorporate nonverbal expressions can influence cultures increasingly reliant on algorithmic information processing; future intelligence and ethics are founded on the choices we now make in what is recognized as valuable data.