Cognitve psychology and HTM Theory


#1

Hi,

I am new on this forum and I looked briefly across the forum. I saw that many people (if not all) are focused on biological aspect of the brain but nobody pay interest on cognitive psychology. When I say cognitive psychology I mean the portion of it that not only try to build theories on how we (or animals) process information but also try to find evidences of their theory in the brain with, for example, fMRI or EEG.

So, my question is : Are there people interested in Cognitive psychology working on HTM Theory and, If not, why ?

Philippe


#2

Cognitive psychology is the main interest in my life as an independent scholar in artificial intelligence. Although my involvement with Numenta goes back to writing a Slashdot story in 2005, I have only recently joined the HTM Forum in order to cross-fertilize ideas between my AI project and the Numenta project. I find the HTM Theory difficult to understand, because it deals with primitive and complex functions of neurons. When I began theorizing on how the mind works as a college sophomore, I was designing neuronal memory channels and motor initiators in a totally top-down approach to the problem of building an artificial Mind. Over the years since 1993 I have built six artificial Minds that think in English, German and Russian. Now I am trying to give my software away to Russian AI programmers who I hope will improve upon my own humble efforts.


#3

Hi Philip,
Welcome. This forum was initially created by Numenta and is moderated by Matthew Taylor, a Numenta employee. Numenta’s goal is to reverse engineer the neocortex.That is, by definition, a very detailed biological endeavor. We want to understand how the neocortex works at a level of detail that is testable by experimental neuroscientists, and also at a level of detail sufficient to create SW and HW technologies that work on the same principles as the neocortex. It is our belief that to create truly intelligent machines we first need to understand what the brain does and how it works.

Psychology and neuroscience are ultimately concerned with similar things such as thought and mental states, but differ greatly in their approach. HTM theory is a theory of how the neocortex works, which means how the detailed biology works. At Numenta we review the literature on EEG and fMRI, but we view anatomy and physiology as more instructive to our goals.

I imagine there are probably other forums that focus on cognitive psychology.
Jeff


#4

Yes, but the first try to explain how we process information of the external world and the second how we process our internal representation. The first can be a great source of inspiration and can help to identify how to define the internal processes and architecture. Furthermore, if the internal “system” is correct, HTM should exhibit some properties from cognitive psychology at the basic level (at the beginning) and at higher levels as you progress, no ?

That is a point of view to see anatomy and physiology as more instructive to your goals but it seems strange to me to agree that there is a link between psychology and neuroscience and to ignore the first one.

Philippe


#5

I think psychology sometimes prompts new topics of focus. For example, the what and where pathways might’ve prompted some of the new research, so maybe it sometimes plays a critical role in research direction. But once HTM researchers know what ability is needed, they need to create a specific algorithmic solution, and that takes a long period of focus on biology. Psychology is more abstract, so it’s harder to convert into specific mechanisms.

HTM might show more and more aspects of psychology, but the brain has a lot of mechanisms HTM won’t need, like emotion and specializations, so it’s hard to say. Right now, it’s hard to judge HTM by psychology, because it’s incomplete and cannot do abstract thought.


#6

Sometimes it is right but it is not always the case, some researchers in experimental cognitive psychology work at very low level and give details for an algorithmic implementation of their theory, especially in recent papers may be because of the new popularity of AI.

It depends on what you want to do with your AI. When you have a specific research goal, you don’t know beforehand all you need and sometimes you find very interesting point of view in subjects that seem slightly irrelevant at start. I prefer to keep an open mind and use all available data.

I won’t argue you with that :slight_smile:

Philippe.