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. The serialization of spreading activation described in Grangers paper and on this list represents what in Novamente would appear to label as an ambient activation process, because it is largely bottom up. I personally found the papers description of the Basil Ganglia quite interesting. The basil ganglia receive a lot of midlevel-and-above-up activation from the prefrontal and motor lobes. Many believe that at least within each of multiple different channels (which may be hierarchical, with some having control over others) it performs some sort of go-no-go and/or winner-take-all or K-winner-take-all) action selections function. This relates not only to motor functions but also to mental functions. So, as opposed to the ambient activation process described above, the Basil Ganglia may represent a more top-down goal-and/or-value-driven attention control process. In fact, for about my last hour at the Similarity Summit 2007 I had one of those "really fascinating conversations" with Todd Huffman ( <mailto:[EMAIL PROTECTED]> [EMAIL PROTECTED]), a brain science type. Near the end of our time together I told him I had developed a hunch that the basil ganglia could select a synchronicity signal of interest, say that being currently received by a node of emotional or goal-related importance. Then -- through the intralaminar nuclei which appears to be influenced by the basil-ganglia (and which are know to be essential to consciousness) -- the cortico-thalamic feedback loop could give preferential dis-inhibition to the individual cortico-thalamic loops of cortical columns from almost any part of the cortex that were firings in sync with the selected synchronicity. Both he and his female friend said that what I had guessed was a known, current hypothesis in brain science. This would, in effect, let the brain tune in on what was exciting that node of importance. In effect, providing a way to do current context-specific backward chaining. A somewhat similar process could be used for forward chaining, and for giving preferential activation to a whole ensemble of activations distributed throughout different parts of the cortex. If there were some mechanism for back propagating the synchronicity used for such selective tungin -- such as through the reciprocal links that exist at almost every level in the cortex and/or by the action of the synchronicity favoring nature of the cortico-thalamic feedback loop, itself, combined with the ability of synchronicity signals to represent a changing mix of multiple different multiplexed signals-- then such context specific backward chaining could spread through multiple activation hops. Since synchronicity is often observed at multiple relatively distant parts of the brain at once, there is reason to believe this might be possible. (Although there are other models for synchronicity spreading, such as those involving rippling waves of dis-inhibitation in the reticular thalamic nuclei, which might, conceiveable, be activated by the Granger's serialzed activations.) In any case, if I dont get jumped on again for daring to share thoughts that are outside the box of the proven and/or totally thought-out, I think Grangers paper also providing useful information, at least to me, about the basil ganglia. I value such information because the basil ganglia appears to be part of the machinery that provides top down control for the cortex and plays an important role in goal, sub-goal, and conscious attention selection. Edward W. Porter Porter & Associates 24 String Bridge S12 Exeter, NH 03833 (617) 494-1722 Fax (617) 494-1822 [EMAIL PROTECTED] -----Original Message----- From: Benjamin Goertzel [mailto:[EMAIL PROTECTED] Sent: Sunday, October 21, 2007 3:05 PM To: [email protected] Subject: Re: Bogus Neuroscience [WAS Re: [agi] Human memory and number of synapses] Loosemore wrote: Edward If I were you, I would not get too excited about this paper, nor others of this sort (see, e.g. Granger's other general brain-engineering paper at http://www.dartmouth.edu/~rhg/pubs/RHGai50.pdf). This kind of research comes pretty close to something that deserves to be called "bogus neuroscience" -- very dense publication, full of neuroanatomic detail, with occasional assertions that a particular circuit or brain structure corresponds to a cognitive function. Only problem: the statements about neuroanatomy are at the [Experienced Researcher] level, while the statements about cognitive functions are at the [First Year Psychology Student Who Took One Class In Cog Psy And Thinks They Know Everything] level. The statements about cognitive functions are embarrassing in their naivete. Richard, I think you do have a point, but as often, I think you overstate it ;-) The title of one of Granger's other papers makes an interesting point: Granger R (2006) Engines of the Brain: The computational instruction set of human cognition. AI Magazine (In press) Let's suppose that he is right and he has found, in some moderately accurate metaphorical sense, "the computational instruction set of human cognition." It's not really clear what this means.... For instance, let's suppose that Susan Greenfield is roughly right -- and concepts, when they rise to attention, take the form of transient neural assemblies, each one of which is assembled based on a core of complexly interconnected neurons. Then, the most Granger's "instruction set" would explain would be some of the mechanics by which these transient neural assemblies form. He refers to olfaction a lot, but Walter Freeman showed years ago that rabbit olfactory cortex is full of complex strange attractors that play a role in olfactory recognition. Most likely similar complex strange attractors (and associated strange transients, associated with Greenfieldian transient assemblies) play a role in cognitive cortex ... but Granger's work tells you none of this. At best it tells you the low-level neural structures and operations that mediate the emergence of these complex dynamics... So, when Granger talks about language learning and language processing, yeah, he seems to be WAY oversimplifying things. Maybe the mechanisms he isolates really ARE in some sense the basic operations underlying linguistic facility, but surely not in the simplistic sort of way he alludes to. Rather, if he's right, it would most likely be because the mechanisms he isolates serve as the infrastructure for some complex dynamical process giving rise to the strange transient assemblies representing linguistic concepts and structures. But then there are a couple missing links, -- explain how Granger's mechanisms or something analogus gives rise to Greenfieldian strange transients, with Freeman-like strange-attractor aspects -- explain how this Greenfield/Freeman stuff can give rise to complex behaviors like language learning In some chapters in my books Chaotic Logic and From Complexity to Creativity, in the late 1990's, I attempted to explain the latter, but didn't finish the job as I got distracted with AI ;-) Basically, one can look at a strange attractor and model its dynamics using formal grammar theory. So, grammars can emerge from complex dynamical systems. This is a means via which symbolic systems can palpably emerge from subsymbolic systems. In physics it's called "symbolic dynamics." Anyway I'm digressing too much into my own weird brain theories (which btw are only loosely connected to Novamente) -- my point is that SOME additional theories like this are necessary to connect Granger's neural ideas to cognition .. you can't just hack them together with glib verbiage as he seems to do in some passages in his papers... OTOH I find his discussion of various issues in neuroscience quite insightful... -- Ben G _____ 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=56155645-f1dc08
