Toward a Non-Reductive Account of Perception
- School of Computing, Queen's
- Department of Psychology, Queen's
Talks at this conferenceAdd a talk
Submission Deadline: Dec. 15, 2022
Decisions Announced: Jan. 30, 2023
Workshop Dates: April or May 2023 (Friday — Monday inclusive); precise dates TBA
Venue: Kingston, ON, Canada
Organizing Committee: Nancy Salay, Department of Philosophy and School of Computing, Queen’s University; Catherine Stinson, Department of Philosophy and School of Computing, Queen’s University
Sponsors: Queen’s University: Department of Psychology; School of Computing; Department of Philosophy;
Workshop Overview: The workshop will be held over four days on Wolfe Island, which offers a distraction-free setting for participants. The small group of cross-disciplinary attendees will include researchers who have been thinking deeply about the problem of perception from both reductive and non-reductive vantage points. The session will serve as the starting point for a series of collaborative papers by the workshop attendees with an aim to publication in a special issue or collected volume. The workshop component will not include an online option but the morning presentations, which are open to all (see below), will.
Theme: How we understand the mechanism of perception guides a wide range of research in areas such as cognitive psychology, neuroscience, AI, philosophy of mind, systems biology, comparative psychology, and neuro-linguistics. This list is not exhaustive. That we have a theory of perception and further that we have it more or less right is thus critical since not doing so will have a negative domino effect on any research that depends on it.
The current mainstream view, however, is a robust, comprehensive account of sensation that stops just short of explaining perception: myriad low level neural processes represent low level sensory features of perceived objects — e.g., pixels, edges, molecules —while deeper neural processes combine these into object representations at increasing levels of granularity — e.g., cat, table, freshly baked bread. To perceive an object on this view is to gain personal level access, in a way to be explained, to these object representations.
An important challenge to this reductive, representational view is the charge that treating perception as a representational process is itself a deep mistake. Anti-representational alternatives have inspired increasingly “embodied” adjustments to the standard view as well as alternative, non-representational theories of perception. Of these the sensorimotor view stands out as a substantive contender, but it too is ultimately a reductive, sensory account that does not address the key personal level features of perception: it does not explain how object wholes become significant to perceivers, how personal level perceptual engagement drives attention, how the awareness of a meaningful, actionable environment plays a causal role in the actions of perceiving agents.
What is meant by “personal level access” and what, if any, functional role it plays in perception are questions that take on a new sense of urgency as AI systems such as autonomous vehicles begin to enter the public space. Determining the extent to which they can be safely integrated will require assessment of the “perception” mechanisms on which they depend but, without a complete theory of perception, we cannot do this, at least not with confidence.
The task of this workshop is to recruit a discipline-diverse array of relevant experts to begin the work of filling in these explanatory gaps. More specifically, the four-day workshop will spearhead a cross-disciplinary collaboration on the following key issues in perception with an aim to developing testable hypotheses that complement and complete our current best view:
— What role does the sensory system play in perception?
— What role do mechanisms of holistic attention play in perception?
— What are the neural markers of holistic attention?
— How do meaningful, actionable landscapes develop for perceivers? What role, if any, does awareness play in this process?
— Which concepts — information-processing, sensorimotor, or a proposed new set — best capture the relation between sensation and perception?
— Which aspects of perception are amenable to reductive explanation and which are not?
— Can dynamic systems concepts be used to develop a non-reductive complement to the best current reductive account of perception?
— Can a non-reductive account of perception shed light on the object recognition problem in AI?
Tentative Schedule: Mornings: participant presentations of research (30-minute with 20-minute Q&A); open to attendance by students and faculty of local departments. Afternoons: work sessions, both smaller group and whole group, focused on developing positions on specific aspects of the problem; open to workshop participants. Evenings: dinner and time for socialising and informal discussion.
Call for Abstracts: The organising committee invites abstract submissions that include the following: 1) Description of applicant contribution to the workshop; 2) Theme questions (see above) that especially relate to applicant’s research and interest; 3) Three relevant publications.
Abstracts should be approximately 500 words. Please submit your abstract in .pdf format to [email protected] with subject heading "PW: SUBMISSION." In the body of the email, please include your name, university affiliation, and presentation title.
If you are interested in registering for the morning presentations online, send an email with the subject heading "PW: ONLINE PRESENTATION REGISTRATION."
Please contact [email protected] with any inquiries.
March 10, 2023, 9:00am EST