Generalization in Mind and Machine
Bowers, J.S (2017-2022)
Funded under: H2020 | ERC-2016-ADG | Advanced Grant, 2.5 million Euros [PI]
Project Website: Generalization in Mind and Machine
Cordis Website: Cordis EU Research Results
OpenAire Website: M and M Generalization in Mind and Machine
Data Management Plan (DMP): Download in .pdf via DMPOnline
Objective: Is the human mind a symbolic computational device? This issue was at the core Chomsky’s critique of Skinner in the 1960s, and motivated the debates regarding Parallel Distributed Processing models developed in the 1980s. The recent successes of “deep” networks make this issue topical for psychology and neuroscience, and it raises the question of whether symbols are needed for artificial intelligence more generally.
One of the innovations of the current project is to identify simple empirical phenomena that will serve a critical test-bed for both symbolic and non-symbolic neural networks. In order to make substantial progress on this issue a series of empirical and computational investigations are organised as follows. First, studies focus on tasks that, according to proponents of symbolic systems, require symbols for the sake of generalisation. Accordingly, if non-symbolic networks succeed, it would undermine one of the main motivations for symbolic systems. Second, studies focus on generalisation in tasks in which human performance is well characterised. Accordingly, the research will provide important constraints for theories of cognition across a range of domains, including vision, memory, and reasoning. Third, studies develop new learning algorithms designed to make symbolic systems biologically plausible. One of the reasons why symbolic networks are often dismissed is the claim that they are not as biologically plausible as non-symbolic models. This last ambition is the most high-risk but also potentially the most important: Introducing new computational principles may fundamentally advance our understanding of how the brain learns and computes, and furthermore, these principles may increase the computational powers of networks in ways that are important for engineering and artificial intelligence.
Bowers, J.S. & Davis, C. J. (2016-2019)
When and Why do Neural Networks Learn Selective Codes?
The Leverhulme Trust. ~£250K
Abstract: This four-year project aims to understand why some neurons respond to information in a highly selective manner. It comprises a series of four computer simulation projects that characterize the representations learned by artificial neural networks across a range of conditions. The findings will demonstrate when and why selective codes are learned in artificial networks and provide insights into why and when neurons respond in this way. A better understanding of the computational advantages of selective coding is fundamental to understanding how the brain supports cognition and will have important implications for theories in psychology and neuroscience.
Davis, C.J. & Bowers, J.S. (2015-2018)
Improving Literacy Outcomes in Struggling Readers: A Randomised Control Study
The Nuffield Foundation, ~£200K
Abstract: This project is a randomised control trial (RCT) to test the efficacy of a morphological training intervention called Structured Word Inquiry (Bowers & Kirby, 2010). This form of instruction makes use of graphical methods that highlight the logic and structure of English spelling, and capitalises on the benefits associated with structuring and organising learning material.
While there are many anecdotal reports of the efficacy of this method, there is little systematic evaluation. However, Structured Word Inquiry may be especially beneficial for children who are resistant to intensive phonics instruction (perhaps due to phonological deficits), because it takes a compensatory approach that focuses on training their stronger skills.
Bowers, J.S. (2011-2013)
The Role of Local and Symbolic Representations in Mind and Brain.
The Leverhulme Trust, ~£200K [PI]
To investigate the nature of mental representations that support word identification and short-term memory using behavioral methods and computer simulations.
Bowers, J.S. (2007-2010)
Bristol site co-PI for “Sound to Sense” (S2S), a Marie Curie Research Training Network (RTN) involving engineers, computer scientists, psychologists, and linguists (£2M).
S2S supports 18 PhD & postdoctoral researchers, as well as workshops on the topic of speech recognition by humans and machines.
Bowers, J.S. & Mattys, S. (2005-2008)
The impact of Early Language Experience on Later Speech Perception
Economic & Social Research Council (ESRC), £152,635 [PI]
Davis, C.J., Bowers, J.S., & Mattys, S. (2006-2007).
Is Speech Perception Influenced by top-down Feedback?
Economic & Social Research Council (ESRC), £44,175 [co-PI].
Damian, M.F., & Bowers, J.S. (2004-2005).
The effects of orthography on language production.
Economic & Social Research Council (ESRC), £41,222 [co-PI]
Bowers, J.S., and Damian, M.F. (2002-2005)
Contrasting two frameworks of connectionism when applied to word identification.
Biotechnology and Biological Science Research Council (BBSRC) £184,400 [PI]
Bowers, J.S. (1999).
Visual Word Recognition in the Left and Right Hemispheres
Economic & Social Research Council (ESRC), £33,661 [PI]