Hardalupas, Mahi
(2021)
How neural is a neural net? Bio-inspired computational models and their impact on the multiple realization debate.
Doctoral Dissertation, University of Pittsburgh.
(Unpublished)
This is the latest version of this item.
Abstract
My dissertation introduces a new account of multiple realization called ‘engineered multiple realization’ and applies it to cases of artificial intelligence research in computational neuroscience. Multiple realization has had an illustrious philosophical history, where multiple realization is when a higher-level (psychological) kind can be realized by several different lower-level (physical) kinds. There are two threads in the multiple realization literature: one situated in philosophy of mind and the other in philosophy of science. In philosophy of mind, multiple realization is typically seen as arbitrating a debate between metaphysical accounts of the mind, namely functionalism and identity theory. Philosophers of science look to how multiple realization is connected to scientific practice, but many question what it is useful for outside of philosophy of mind.
My dissertation addresses this gap by drawing on cases from machine learning and computational neuroscience to show there is a useful form of multiple realization based on engineering practice. It differs from previous accounts in three ways. First, it reintroduces the link between engineering and multiple realization, which has been mostly neglected in current debates. Second, it is explicitly perspectival, where what counts as multiple realization depends on your perspective. Third, it locates the utility of engineered multiple realization in its ability to support constraint-based reasoning in science. This provides an answer to concerns about the utility of multiple realization in the philosophy of science literature and explains how deep neural networks can provide understanding of the brain. The first half of this dissertation proposes my account of Engineered Multiple Realization and applies it to scientific cases. The second half considers implications and connections to the modelling literature.
Share
| Citation/Export: |
|
| Social Networking: |
|
Details
| Item Type: |
University of Pittsburgh ETD
|
| Status: |
Unpublished |
| Creators/Authors: |
|
| ETD Committee: |
|
| Date: |
3 May 2021 |
| Date Type: |
Publication |
| Defense Date: |
6 April 2021 |
| Approval Date: |
3 May 2021 |
| Submission Date: |
8 April 2021 |
| Access Restriction: |
No restriction; Release the ETD for access worldwide immediately. |
| Number of Pages: |
140 |
| Institution: |
University of Pittsburgh |
| Schools and Programs: |
Dietrich School of Arts and Sciences > History and Philosophy of Science |
| Degree: |
PhD - Doctor of Philosophy |
| Thesis Type: |
Doctoral Dissertation |
| Refereed: |
Yes |
| Uncontrolled Keywords: |
history and philosophy of science, multiple realization, bio-inspiration, deep learning, computational neuroscience |
| Date Deposited: |
03 May 2021 15:12 |
| Last Modified: |
03 May 2021 15:12 |
| URI: |
http://d-scholarship.pitt.edu/id/eprint/40634 |
Available Versions of this Item
Metrics
Monthly Views for the past 3 years
Plum Analytics
Actions (login required)
 |
View Item |
![[feed]](/images/feed-icon-32x32.png){kind=icon}