Thursday, April 17, 2014

Foundations of Mind Workshop series, UC Berkeley

Foundations of Mind Workshop series, UC Berkeley

The Foundations of Mind (FOM) project believes that it is possible that, far from showing incremental (let alone as is claimed, exponential) progress, modern neuroscience is a failed paradigm insofar as it focuses on the attempt to explain mentation in neural terms. Moreover, the reduction of cognitive science to neuroscience has worsened matters. Finally, even correcting these trends may leave a sphere of human action beyond the explanatory scope of science. On Mar 6-7 we ran a successful conference (see which has begotten an active community of researchers who are now running a fortnightly seminar series, Fridays Beach room Tolman Hall UC Berkeley.

Wrt the first point, we do not have a single example of a sentence or other meaningful symbol being understood by an attested neural process and seem, if anything, to be going ever further from such a goal as the models get ever more primitive after the towering work of the 1980's in neural nets. Secondly, the assumption that mind can be described as a set of faculties, each with an anatomically distinct location, was regarded as risible phrenology until its 1990's comeback as fmri. Thirdly, fmri is a scalar formalism, famously prone to fraud, at a time when it is increasingly clear that the brain not only uses vectors, but that non-linearity is deep in its working ethos.

The reduction of cognitive science to neuroscience has combined with a mania for “Big data” to elide the contribution of linguists. The result has been machine translation systems that are dangerous in their inaccuracy and steadfast in their refusal to allow syntax, semantics and pragmatics to constrain the search space. Even if they did, it is unlikely that the hope for machines to “understand” language is anything but a negatively eschatological aspiration.

That brings us to our final point; in the classical Von Neumann interpretation of quantum mechanics, decision and observation is in the classical realm, outside the purview of the system. Can it be the case that modern neuroscience has been incorrect in failing to give this proper weight? Is it possible that, even after we as responsible scientists get our formalisms right, there will remain limitations in our simulation of behavior by machines?

All seminars are free and open to the public.

Workshop in Beach Room, 3rd floor Tolman hall, UC Berkeley April 18 2014 2pm
2-10 pm Introductory remarks Seán O Nualláin
2-15pm Henry Stapp on “Quantum Mechanics and neuroscience ”

3-15 pm Stan Klein Connecting visual qualia to their neural correlates”

How does simulation of a single neuron in ones visual system produce the richness of subjective color perception? This is the topic of connecting qualia to neural activity. The capability for achieving that connection has only become available in the past six months, because of a new instrument developed at UC Berkeley. But before getting to the recent experiments some context is appropriate in the context of this 20th anniversary of the Tucson "Toward a Science of Consciousness" (TSC) meetings. Who can forget Chalmer's presentation of the "Hard Problem" at TSCII in 1996. The 764 page book of papers from that meeting is a treasure and is unequaled in capturing the state of the field at that point. The challenge of connecting the hard problem of qualia to the neural correlates of the brain is still with us and I'll argue that advances in technology may enable substantial progress.  My article in the Vision and Consciousness section of the TSCII book was titled "Double Judgment Psychophysics for Research on Consciousness: Applications to Blindsight" ( My present title could have been the same as the old title but replacing "Double" to "Triple", and "Blindsight" to "color" as I now discuss.
A new instrument that enables single neuron brain stimulation has been developed by Austin Roorda of UC Berkeley. The new technology that combines adaptive optics and super precise image stabilization enables one to do careful psychophysics on individual cones and ganglion cells. Retinal ganglion cells (RGC) are special since they are the bottleneck for vision, carrying information to the brain along the optic nerve. Roorda's lab is next to mine and we have been developing new approaches for single and double RGC stimulation.  One of the big surprises is that activation of single retinal neurons can produce a wide variety of color percepts. It had previously been thought that single RGC stimulation would produce limited color percepts (red, green, yellow, blue) based on the opponent color mechanisms found in RGCs and the brain's lateral geniculate nucleus (LGN). But a much greater diversity of colors were found. In our experiments for every single RGC stimulation we make a triple judgment on the perceived hue, saturation and intensity of the stimulus. Several competing hypotheses regarding the mechanisms that produce the surprisingly diverse color percepts are now being tested and will be discussed. The new instrument opens up a great variety of experiments linking neural activity to perception. 

4pm Gautam Agarwal, et al.
The emergence of information in mesoscopic measures of brain activity”

4-45 pm
Seán O Nualláin of the
The great neuroscientist Karl Pribram and equally great physicist David Bohm collaborated a generation ago with, among others, Jiddah Krishnamrti to produce a completely new worldview. It provided a narrative, based on cutting-edge science on how the evolutionary process allowed nature to express itself as the “implicate” becoming “explicate” and presenting itself to consciousness manifest in us. This beautiful paradigm eventually failed to overcome resistance centering on the neurophysiological plausibility and lack of detail.

In 1999, after having met Seán O Nualláin at a conference on Gurdjieff organized by Seán’s close friend Jacob Needleman, Karl Pribram spent a week at the O Nualláin lab in Dublin. There, for the first time, a detailed computational implementation of the Pribram / Bohm work was done. Far from being merely a beautiful theoretical framework, it was found that the view of dendrodendritic connections between neurons as the critical computational operation in the brain explained many heretofore intractable problems about sense perception. Moreover, it was consistent with Stapp's work on quantum neural activity qua harmonic oscillators.
May 2 2-30 pm Walter Freeman on neural dynamics at the mesoscopic and macroscopic levels
4-30 pm The microscopic level Seán O Nualláin

“Resonate and fire; how non-spiking neurons process sensory data”

While it is universally accepted that several types of cells in the
retina, including rods, do not spike, little investigation has been
done to ask how sense-data can be processed without neural firing.
Moreover, the fact that timing of afferent impulses to a neuron shapes
its firing has not received enough attention. This talk argues that
firing is actually a limit case of a more general mode of neural
functioning in which subthreshold oscillations have a critical causal
role. A harmonic oscillator model of the neuron is sketched, beginning
from a compartmental Hodgkin-Huxley model, and it is shown how the
emergent principles operate equally for tactile, auditory, and visual

The talk ends with speculation on the role of attention and consciousness.

Finally, please note that this has been sent out with the support, indeed the encouragement of various list moderators; if you do not want to receive more, please mark as spam.

No comments:

Post a Comment