advancements in connectionism

I have chosen connectionism and its potential ability to model various learning processes in the brain by a multi-disciplinary approach that combines many different theoretical approaches that have recently been given a big boost with advances in technology. The basic principles that define the connectionism model involve a sense of biological realism that combines interconnected networks that form a more complex network that could explain the processes within the human brain, as well as also serve as a model that could also be used to develop non-human networks such as AI for instance. Although it is not necessarily clear how this research might be relevant to my career goals at present, the research seems to be developing so fast that knowledge of this subject could be entirely relevant within the next five years.

Dalege, J., van den Berg, H., Borsboom, D., Conner, M., & van der Mas, H. (2016). Toward a Formalized Account of Attitudes: The Causal Attitude Network (CAN) Model. Psychological Review, 2-21.

Mayor, J., Gomez, P., Chang, F., & Lupyan, G. (2014). Connectionism coming of age: legacy and future challenges. Frontiers in Psychology. doi:10.3389/fpsyg.2014.00187

Nelson, R. (2013). Expanding the Role of Connectionism in SLA theory. Language Learning, 1-33.

Plaut, D., & Vande Velde, A. (2017). Statistical Learning of Parts and Wholes: A Neural Network Approach. Journal of Experimental Psychology, 318-336.

Wiltshire, T. (2015). A Prospective Framework for the Design of Ideal Artificial Moral Agents: Insights from the Science of Heroism in Human. Minds and Machines, 57-71.

This article was chosen because it provides a rich and detailed history of how this model developed, as well as some discussion of where the model could go in terms of advancement into the future. In 1986, Rumelhart and McClelland took the cognitive science community by storm with the Parallel Distributed Processing (PDP) framework; which sought to construct at the algorithmic level models of cognition that were compatible with their implementation in the biological substrate (Mayor, Gomez, Chang, & Lupyan, 2014). After walking through some of the obsticles that the theory has so far embraced, it talks about its key challenge, learning abstract structural representations, and how there are many gaps that need to be filled before this model could explain complex intelligence.

Dalege, J., van den Berg, H., Borsboom, D., Conner, M., & van der Mas, H. (2016). Toward a Formalized Account of Attitudes: The Causal Attitude Network (CAN) Model. Psychological Review, 2-21.

This article evaluates the possibility of the CAN model to explain the reactions that people have to events, as well as the interactions between these interactions. For example, when a person jumps at the sight of a snake, they are not undergoing a rational discussion of the threat of the snake in their head, rather, they just react based on their "attitude" towards snakes. Therefore, understanding how these attitudes are structurally represented in the neural networks in the brain can help to begin to understand how such reactions manifest from a connectionism perspective and potentially lead to further study.

Plaut, D., & Vande Velde, A. (2017). Statistical Learning of Parts and Wholes: A Neural Network Approach. Journal of Experimental Psychology, 318-336.

This article looks at the phenomenon of statistical learning and how insights from statistical learning findings might be able to help identify some of the properties that neural networks must undergo to learn various materials. For example, learners are sensitive to different "chuckings" of information from different modalities, including auditory and visual sources. The researchers build a Bayesian statistical model to see if insights can be gained to help advance some of the problematic obstacles in understanding neural networks.

Nelson, R. (2013). Expanding the Role of Connectionism in SLA theory. Language Learning, 1-33.

This article uses the ability to learn a second language, second language acquisition (SLA), in an attempt to show how bilingual network models could be reduced to a more biologically realistic connectionism model that could offer more insights into how multiple networks coalesce. A second language does not always take the same learning skills as a first language if not learned as a child. Yet, despite the ways in which these languages are learned, there is likely some overlap in the neural networks that support them. By teasing out data from samples of bilingual and monolingual populations, researchers stand the potential to unlock insights about such connections.

Wiltshire, T. (2015). A Prospective Framework for the Design of Ideal Artificial Moral Agents: Insights from the Science of Heroism in Human. Minds and Machines, 57-71.

This article tackles the moral aspects of building AI that has some concept of morality built into whatever composition of neural networks goes into creating advanced AI systems. Although this article tackles the connectionism model from a theoretical perspective that is looking specifically at its future capabilities, it is interesting to start thinking about the possibilities that advancements in connectionism could create and how to start planning for them beforehand. This article lays out some ideas about how heroism could possibly integrated into these systems to form some form of moral basis for behavior.

Summary

The five articles chosen all come from significantly different research perspectives and show just how exciting research in this field has become, evident by its popularity. Many researchers are trying to use existing models and apply them from the connectionism perspective to try to unlock the secrets of how these networks may relate to each other. From my perception, although many of the technical aspects are hard to follow, it is clear to see how this model rests on one of the frontiers of the next generation of scientific breakthroughs.

Works Cited

Dalege, J., van den Berg, H., Borsboom, D., Conner, M., & van der Mas, H. (2016). Toward a Formalized Account of Attitudes: The Causal Attitude Network (CAN) Model. Psychological Review, 2-21.

Mayor, J., Gomez, P., Chang, F., & Lupyan, G. (2014). Connectionism coming of age: legacy and future challenges. Frontiers in Psychology. doi:10.3389/fpsyg.2014.00187

Nelson, R. (2013). Expanding the Role of Connectionism in SLA theory. Language Learning, 1-33.

Plaut, D., & Vande Velde, A. (2017). Statistical Learning of Parts and Wholes: A Neural Network Approach. Journal of Experimental Psychology, 318-336.

Wiltshire, T. (2015). A Prospective Framework for the Design of Ideal Artificial Moral Agents: Insights from the Science of Heroism in Human. Minds and Machines, 57-71.