Georges Rey’s book “Contemporary Philosophy of Mind” is mainly a description of his theory of how functionalism deals with the higher level functional organization of the mind. To what extent are philosophers like Rey made useless by the limitations of “encodingism”? It seems that Rey is willing to accept the reality that functionalistic accounts of mind should be “grounded” or “naturalized” by including mechanisms for getting representational content into their models of mind, but he, like most functionalists seems to be satisfied to let semantics enter into their models of mind by way of the users of those models. Rey suggests that he could live with some type of neural network-based mechanism for getting new representational content into his model of mind, although he is not really waiting impatiently for such a source of representational content. Similarly, I suspect that most connectionists and classical top-down artificial intelligence researchers (use Marvin Minsky as the prototype) assume that human programmers are their preferred means of inserting representational content into their programs.

Rather than try to get people to switch from encodingism to interactivism, I am almost willing to “write-off” all of the functionalists and AI researchers who want to continue avoiding the problem of how systems can autonomously learn representational content. My attitude is, let them play their game, while it will be left to others to confront the study of how to naturalize representational content. If it is the case that only a minority of people concerned with mind stand prepared to confront this issue, should we make an effort to coordinate their efforts? I am particularly concerned that people who are working from higher levels of functional organization and people working upwards from the low levels of the mind/brain hierarchy are not using the same language and not communicating effectively. Can we recognize this problem and deal with it so as to help maximize the effectiveness of all of our efforts?

Some other people (besides Edelman) came to my mind as I read Bickhard. Dug Hofstadter’s (Godel, Echer, Bach) deals well with hierarchical organization, self-representation, and self-replication.  Also, Bickhard’s discussion of adaptive systems reminds me of Stu Kauffman (The origins of order) and the rest of the Santa Fe “complex adaptive systems” group. I have long felt that a weakness of Edelman’s Theory of Neuronal Group Selection is that he did not attempt to link his theory to the work of others that is centered on finding useful ways to deal with the control and organization of complex adaptive systems. I do not fault Edelman for this, since the new science of complexity was still being formed when he created his theory. I am pleased to see Bickhard integrate control theory into his idea of interactive representation. Can we all agree on a common language for dealing with these issues? If Bickhard says,

“differential functional indications in the context of a goal directed system constitute representation”

while Edelman says,

“a conceptually based memory system correlates the interactions between categorized exteroceptive signals and interoceptive signals that reflect homeostatic needs”

and Kauffman says,

“learning is a walk in synaptic weight space seeking attractors that satisfy constraints imposed by certain essential internal variables”

but everyone is talking about the same thing, then we have a real communication problem. This really bothers me. Does it bother you?

Bickhard, Edelman, and I seem to be is basic agreement about how brains obtain meaningful representational content through the interaction of organisms with their environments.

As a working molecular neurobiologist I am still in the mode of identifying the key molecular components of synapses. Once we have the synapse components in the bag, we will then be able to tackle the task of figuring out the interactions of those components which will allow us to start climbing up the hierarchy of brain organization to the level of neural networks. It is my conviction that the over-simplified view of neurons and synapse function that connectionism has been built on is the main reason why connectionists are stuck with computer programs that are so poor at the task of unsupervised learning. We really need to know more about synapses.

Even though the task of characterizing the key components of synapses continues, it is not too early to start incorporating more complex synapse models into models of neural networks. Thus, I have an amateur’s interest in higher levels of functional organization and have done a small amount of neural network modeling in my “spare” time. My approach is entirely along the lines of Edelman’s work, although I have also been heavily influenced by people like Francis Crick, Terry Sejnowski, Dug Hofstadter, Stu Kauffman, and Eric Kandel. As is the case in Edelman’s work, the approach is to try to take the best available information from neuroanatomy, behavioral studies, electrophysiology, and molecular studies and combine them into a biologically plausible network model that is itself part of a model of an autonomous organism interacting with a complex environment. The basic idea of the relationship between analysis of the molecules of synapses and higher order brain functions is that the molecular analysis is showing us the required level of functional complexity in synapse regulation that brains rely on to make learning and memory possible in neural networks.

I am interested in better relations between:

1) philosophers who are working on mind in a “downward” direction from the higher levels of functional organization

and

2) neurobiologists who are working “upward” towards mind from the lower levels of detailed molecular mechanisms of brain processes.

When I see Bickhard and Edelman in basic agreement about how to avoid the limitations of classical functionalism and connectionism, I am sure that by working towards each other along the levels of hierarchical organization, we can meet in the middle with a satisfactory outcome of consilience (as championed in E. O. Wilson’s recent book of that name).
  

Go to John's Book Page.

Go to John's Home Page.

Psychology Today

Magazine Description The mission of Psychology Today is to take information from the latest research in psychology and make it useful to people in their everyday lives. This bimonthly magazine shows readers how to improve themselves, their families and relationships, and their experiences in the workplace. In addition, the magazine looks at societal issues so that readers can better understand the world in which they live.