Richard, You might be interested to know how much attention one of your articles has gotten in the <mailto:agi@v2.listbox.com> agi@v2.listbox.com mailing list under the RE: Bogus Neuroscience [WAS Re: [agi] Human memory and number of synapses thread, which has been dedicated to it.
Below is a message I sent in defense of your paper. If you have comments to either me or the list I would be interested in hearing them. Edward W. Porter Porter & Associates 24 String Bridge S12 Exeter, NH 03833 (617) 494-1722 Fax (617) 494-1822 [EMAIL PROTECTED] -----Original Message----- From: Edward W. Porter [mailto:[EMAIL PROTECTED] Sent: Monday, October 22, 2007 1:34 PM To: agi@v2.listbox.com Subject: RE: Bogus Neuroscience [WAS Re: [agi] Human memory and number of synapses] Dear Readers of the RE: Bogus Neuroscience Thread, Because I am the one responsible for bringing to the attention of this list the Granger article (Engines of the brain: The computational instruction set of human cognition, by Richard Granger) that has caused the recent kerfuffle, this morning I took the time to do a reasonably careful re-read of it. I originally read it when I was interested in trying to learn about the basil ganglia, and didnt read in depth much beyond its initial description of how it can serialize activations from a set of active nodes and learn patterns from such serial activations. And it was this learning from temporally sequential activations that caused me to cite the paper to Vladmir. I had totally forgotten the articles initial, arguably grossly overreaching claims of its own importance, because that wasnt what I remembered as being important it. Upon my complete re-reading this morning, I think, overall, this paper represent valuable work. I think its actual brain science is interesting and important. Its description of the basil ganglia certainly advanced my knowledge substantially. Its basic message about the cortico-thalamic loops is important that, in general, each cortical columns in the cortex has two types of loops through the thalamus: a core loop that feeds directly back to itself; and a matrix loop that feeds forward to widely distributed portions of the cortex. A significant portion of the paper is based on computer simulations. This makes the resulting observations somewhat questionable, since the accuracy of neural models can vary tremendously. But I welcome the general effect the increasing use of computational neural modeling has had on brain science. It lets us create models much more complex than we ever could in our own human minds, and then give them a spin. I think his notion that the combination of the two loops through the cortex allows spreading activation, particularly that between different localized topological maps, to use a sequential coding to, in effect, bind information is extremely interesting, and potentially valuable. It takes a fair amount of thought to understand the significance of this. Once, when we were both young and single and living in Manhattan I met a woman at a party who worked as an Asian art specialist for one of the worlds biggest art auction houses. I told her I had seen an excellent exhibition of 19th century Japanese art and artifacts and had blown me away with its abstraction and minimalism. She responded that in Japanese literature and art it is often a sign of respect for the intelligence of your readers and viewers to relay your message in as few words or as little detail as possible. In a similar vein, when Granger says his paper describes the basic mental operations from which all complex behavioral and cognitive abilities are constructed I think he assumes his intended readers will be quite intelligent enough, well versed in brain and cognitive science, and willing to take the time to understand the potential implications of what he is saying. I think he assumes that such reader, and to a certain degree further research and though, will fill in much of what is left unsaid. If you think about the sequential grammar he describes and include the ability for time dilation and compression he incorporates from other papers I have not read, it would possibly, in conjunction with prior knowledge, provide a mechanism for the learning, perception ,and recall of compositional structures having all the invariance of Hawkinss hierarchical memory. It not only provides for dealing with compositional patterns that are static, but also ones that are temporal. It also allows patterns to be learned that have elements spanning multiple topological regions of the brain. This is interesting and quite valuable. As I said above, it leaves many things unsaid and unclear. For example, does it activate all or multiple nodes in a cluster together or not? Does it always activate the most general cluster covering a given pattern, or does it use some measure of how well a cluster fits input to select what, and to what degree, cluster(s) in the generalization hierarchy spreads its(their) activation through the matrix loop? Is it correct to assume that this form of sequential spreading activation can take place between massive number of subconsciously activated nodes simultaneously, or is it limited to a relatively few, or near conscious nodes? How exactly does the model of the basil ganglia described in the earlier part of this paper plug into the operation of the core and matrix loops described in its later part. How does it handle sequential activations that are feed to it in a different order than that originally learned. Etc. But, if you assume its basic hypotheses are correct and novel, and if you use your imagination, you can see what a valuable addition to brain science this paper might be. In his Sun 10/21/2007 2:12 PM post Richard Loosemore cited failure to answer the following questions as indications of the papers worthlessness. RICHARD>> How does it cope with the instance/generic distinction? I assume after the most general cluster, or the cluster having the most activation from the current feature set, spreads its activation through the matrix loop, then the cluster most activated by the remaining features spreads activation through the matrix loop. This sequence can continue to presumably any desired level of detail supported by the current set of observed, remembered, or imagined features to be communicated in the brain. The added detail from such a sequence of descriptions would distinguish an instance from a generic description reprsented by just one such description.. RICHARD>> How does it allow top-down processes to operate in the recognition process? I dont think there was anything said about this, but the need for, and presence in the brain of, both top-down and bottom-up processes is so well know as to have properly been assumed. RICHARD>> How are relationships between instances encoded? I assume the readers will understand how it handles temporal relationships (if you add the time dilation and compression mentioned above). Spatial relationships would come from the topology of V1 (but sensed spatial relationships can also be build via a kohonen net SOM with temporal difference of activiation time as the SOMs similarity metric). Similarly, other higher order relationships can be built from patterns in the space of hierarchical gen/comp pats networks derived from inputs in these two basic dimensions of space and time plus in the dimensions defined by other sensory, emotional, and motor inputs. [I consider motor outputs as a type of input]. RICHARD>> How are relationships abstracted? By shared features. He addresses how clusters tend to form automatically. These clusters are abstractions. RICHARD>> How does position-independent recognition occur? He deals with this. His nodes are nodes in a hierarchical memory that provides degrees of position and shape invariance, or the type mentioned by Hawkins and the Serre paper I have cited so many times. Grangers figures 6 and 7 indicates exactly this type of invariance. RICHARD>> What about the main issue that usually devastates any behaviorist-type proposal: patterns to be associated with other patterns are first extracted from the input by some (invisible, unacknowledged) preprocessor, but when the nature of this preprocessor is examined carefully, it turns out that its job is far, far more intelligent than the supposed association engine to which it delivers its goods? What he feeds to his system are things like the output of Gabor filters. I dont think a Gabor filter is something that is far, far, more intelligent than the supposed association engine to which it delivers its goods. This is just an example of how a serious attempt to understand what is good in Grangers paper, and to expand on those good features, overcomes a significant number of the objections raised by those whose major motivation seems to be to dismiss it. Wikipedia, that font of undisputed truth, defines Cognitive science as Cognitive science is most simply defined as the scientific study either of mind or of intelligence (e.g. Luger 1994). It is an interdisciplinary study drawing from relevant fields including psychology, philosophy, neuroscience, linguistics, anthropology, computer science, biology, and physics Based on this definition I would say the cognitive science aspect of Grangers paper, although speculative and far from fully fleshed out, is actually quite good. Ed Porter Edward W. Porter Porter & Associates 24 String Bridge S12 Exeter, NH 03833 (617) 494-1722 Fax (617) 494-1822 [EMAIL PROTECTED] -----Original Message----- From: Richard Loosemore [ <mailto:[EMAIL PROTECTED]> mailto:[EMAIL PROTECTED] Sent: Sunday, October 21, 2007 9:37 PM To: agi@v2.listbox.com Subject: Re: Bogus Neuroscience [WAS Re: [agi] Human memory and number of synapses] Edward W. Porter wrote: > As Ben suggests, clearly Grangers title claims to much. At best the > article suggests what may be some important aspects of the computational > architecture of the human brain, not anything approaching a complete > instruction set. > > But as I implied in my last post to Richard Loosemore, you have to > forgive academics for aggressive marketing, because publish or perish > seems to have replaced by market or perish. But in my time as a cognitive scientist, I have had to read through hundreds (perhaps thousands) of papers in which the author(s) made wild claims that really amounted to nothing more than aggressive marketing designed to further their career. I have had large chunks of my time wasted by this self-aggrandisement. It took me at least five years of struggle to get to the point where I could start to have the confidence to call a spade a spade, and dismiss stuff that looked like rubbish. Now, you say "we have to forgive academics" for doing this? The hell we do. If I see garbage being peddled as if it were science, I will call it garbage. Richard Loosemore. ----- This list is sponsored by AGIRI: <http://www.agiri.org/email> http://www.agiri.org/email To unsubscribe or change your options, please go to: <http://v2.listbox.com/member/?&> http://v2.listbox.com/member/?& _____ This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/? <http://v2.listbox.com/member/?& > & ----- This list is sponsored by AGIRI: http://www.agiri.org/email To unsubscribe or change your options, please go to: http://v2.listbox.com/member/?member_id=8660244&id_secret=56364254-10bd62