CONCEPT CELLS = GRANDMOTHER CELLS
On Mon Aug 8, 2011 at 1:21 AM Asim Roy posted the topic on a very broad list including robots (robot@blueyonder.co.uk) to advertise about the scientific debate and the discovery of “concept cells” Asim Roy:We have been debating single cell recordings and concept cells for over a month now” .....From this “broad” list I’ve picked the topic of discussion where Dr. Itzhak Fried, Christof Koch, Quian Quiroga, Walter Freeman commented the “concept cells” idea
On 
Thu, Aug 11, 2011 at 12:50 PM Dorian Aur :Asim, your idea to add together various scientists with similar interests represents an important step to trigger a debate that can indeed solve current issues regarding data analysis and the nature of neural code.I’ve had myself the privilege to analyze some excellent recordings from Dr Fried lab. In general it is an exception to get four good neurons where spike directivity (SD) can be computed (the electrodes need to be together in a tetrode configuration). Dr Fried I’m really impressed, excellent recordings! Therefore in an attempt of pursuing the truth I have analyzed some of these recordings from a different perspective and took the liberty to post here the outcome.Dorian’s summary of some issues in interpreting temporal coding - includes notes from Dr. Itzhak Fried and Christof Koch, Quian Quiroga, Walter Freeman
1. Work on brain-machine interfaces based on single-neuron activity is quite standard now and being pursued by a number of groups.
Itzhak Fried: I would not describe BMI using single neuron activity as quite standard. Most of the existing data is with frontal/motor or parietal neuron.
Christof: Yes. Most such BMI operate in motor, pre-motor and parietal cortex.
Dorian :There are several issues in interpreting single neuron activity using the firing rate or interspike interval. Several important details are missing in temporal patterns (see http://dx.doi.org./10.1016/j.jneumeth.2006.05.003; http://dx.doi.org./10.1016/j.jneumeth.2005.05.006 and the book Neuroelectrodynamics: Understanding the Brain Language, http://dx.doi.org/10.3233/978-1-60750-473-3-i for a model of computation) which are important in information coding
2. Concept cells are comparable to place cells in rodents (concept cells = place cells) and therefore not a finding that surprises the neuroscience community.
Itzhak Fried: Concept cells are not place cells but I proposed that they can be viewed as "place cells" in a different "attribute or feature space". They do share with place cells coding properties, that is: specificity , invariance, sparseness and the explicit nature of the code. One can speculate that the mechanism developed for coding of space in rodent hippocampus has evolved to accommodate more elaborate abstraction in humans. As for "surprises", it is difficult to surprise the neuroscience community, but for us the explicit nature of the code on the single neuron level was a surprise.
Christof Koch: There are some similarities to place cells in rodents. However, we find these highly selective cells in all regions of the MTL, not just the hippocampus. How far this comparison goes is not clear (what, for example, is the analog of grid cells in the entorhinal cortex?)
Dorian: Different experiments in rats or humans show strong similarities that can provide meaningful explanation for these data. However, an understanding of presented examples cannot solely come from a firing rate analysis. A relevant example shows how neurons operate during a T-maze procedural learning task. (D Aur, and M Jog, Reading the Neural Code: What do Spikes Mean for Behavior?. Available from Nature Precedings <http://dx.doi.org/10.1038/npre.2007.61.1, 2007)
The "expert" neurons in striatum during T-maze learning provide a similar behavior, however they extensively fire only before learning. In order to understand the meaning of their firing activity a different measure was computed and analyzed. Spike directivity is a vector that reflects the distribution of electrical patterns in recorded spikes. During learning these "expert" neurons reduce their firing rate. After one week of training the neurons generate only few spikes between the tone and turn starts. This represents the critical moment when the decision regarding turning is taken. After training these “expert” neurons show less random spike directivities (a preferred direction of AP propagation)than before training . The delivered spikes after the tone predict the turning direction on the T-maze. In many cases the firing rate cannot be estimated (one spike in single trials)
The first counterexample: When it fires the same neuron can code for the left turn or for the right turn depending on the context (high or low tone, the T-maze task, http://neuroelectrodynamics.blogspot.com/p/spike-directivity.html
The second counterexample: The same neuron responds with the same firing rate for two different objects (spider, Jennifer Aniston) the difference occurs in the preferred spike directivity see http://neuroelectrodynamics.blogspot.com/p/spike-directivity.html
The outcome in spike directivity is a counterexample for temporal coding. The spikes cannot be added since they provide different semantics (apples and oranges). That’s the beauty of counterexamples. You only need one single counterexample to throw down a "solid"theoretical construct of temporal coding. No need for other examples that reinforce the temporal coding!
In this case:
(i) During learning several options are explored and the strong firing rate reflects uncertainty in these “expert” neurons. Therefore, we hypothesized that strong firing (with strong variability of spike directivity) represents a way to search for a correct solution during learning
(ii) After learning all these cells provide an efficient response with only few spikes for the same event.
(iii) The reduction in uncertainty generates a meaningful outcome that can be observed in a preferred direction of spike directivity
Therefore, a decrease of uncertainty is reflected in a reduced number of spikes delivered by a cell, an efficient response. Contrary to current belief an increase in the firing rate may show uncertainty, a searching process required to deliver a solution.
Following a similar analysis, the cells from MTL can display a similar behavior. If two different objects are presented they can be separated fast in these neurons since they generate different spike directivity orientations (see http://neuroelectrodynamics.blogspot.com/p/spike-directivity.html).
3. The concept cells were found in different regions. For example, “James Brolin” in right hippocampus, “Venus Williams” in left hippocampus, “Marilyn Monroe” in left parahippocampal cortex, “Michael Jackson” in right amygdala.
Itzhak Fried: Yes, but they may represent different levels of abstraction or invariance in each of these regions. Although they were found in different MTL regions , the highest degree of invariance (across modalities) was in hippocampus and entorhinal cortex . Also remember the latency of the response, usually around 350 msec.
Christof: YesIn this case:
(i) During learning several options are explored and the strong firing rate reflects uncertainty in these “expert” neurons. Therefore, we hypothesized that strong firing (with strong variability of spike directivity) represents a way to search for a correct solution during learning
(ii) After learning all these cells provide an efficient response with only few spikes for the same event.
(iii) The reduction in uncertainty generates a meaningful outcome that can be observed in a preferred direction of spike directivity
Therefore, a decrease of uncertainty is reflected in a reduced number of spikes delivered by a cell, an efficient response. Contrary to current belief an increase in the firing rate may show uncertainty, a searching process required to deliver a solution.
Following a similar analysis, the cells from MTL can display a similar behavior. If two different objects are presented they can be separated fast in these neurons since they generate different spike directivity orientations (see http://neuroelectrodynamics.blogspot.com/p/spike-directivity.html).
3. The concept cells were found in different regions. For example, “James Brolin” in right hippocampus, “Venus Williams” in left hippocampus, “Marilyn Monroe” in left parahippocampal cortex, “Michael Jackson” in right amygdala.
Itzhak Fried: Yes, but they may represent different levels of abstraction or invariance in each of these regions. Although they were found in different MTL regions , the highest degree of invariance (across modalities) was in hippocampus and entorhinal cortex . Also remember the latency of the response, usually around 350 msec.
Dorian :Intracellularly within molecular structure specific information is "read" and "written" in these cells during AP generation . In order to have the concept of "Jennifer Aniston" information from many cells is inferred synaptically and non-synaptically. Therefore, many neurons fire almost simultaneously in different brain regions (MTL hippocampus , entorhinal cortex)
(i) However, a too strong increase of firing rate may show high uncertainty - a searching process required to identify the presented object.
(ii) In order to represent a particular feature associated with a certain presented image (e.g. Jennifer Aniston) these neuron can generate low firing rate with consistent preferred spike directivity
(iii) The semantics do not appear in the firing rate!!! (ISI) therefore statistical analyzes of firing rate (ISI) can be highly irrelevant to determine the meaning of firing. The same neuron can code for different features in different spikes depending on presented context.
4. The sister cells (e.g. other Jennifer Aniston concept cells) are not necessarily in contiguous locations in the brain. They could be in different hemispheres and different regions within a hemisphere. (“The subject most likely activated a large pool of neurons selective to ‘Johnny Cash’ even though the feedback was only based on just one such unit. We identified 8 such units in a total of 7 subjects.”)
Itzhak Fried: The sister cell may be a confusing term, but a major point is that organization of "concept cells" is not columnar or topographic. Given their sparse and nontopographic distribution it would be difficult to trace them on fMRI.
Dorian: There is little information in the temporal code (firing rate, ISI)
Based solely on firing rate (ISI) analyses it is very hard to figure out the role of certain spikes -see the counterexample where the same neuron provides different semantics when it fires http://precedings.nature.com/documents/61/version/1
5. Even though a million cells are activated by an image of Jennifer Aniston, and say 12 of them are Jennifer Aniston concept cells, in your experiments, you tracked only one such concept cell and that was good enough. There was no need to “read out” other Jennifer Aniston concept cells, wherever they were, as would be required in a distributed representation framework.
Itzhak Fried: Yes. But I suspect more than a million cells are activated by Jennifer Aniston and they could probably be arranged on a variance scale with our "concept cells" at the extreme low. Still it is easier to find a concept cell than a Higg's boson.
Christof: We have no idea whether J. Aniston activates a million of such cells. Yes, the movie of the superimposed images was based on four selective units of a presumably much larger pool. It is well possible that if we had recorded from more sister neurons, control would have been swifter or more precise or more reliable.
Dorian: (i)These cells do not fire only for Jennifer Aniston as presented in Quiroga et al., 2005. (ii) In different contexts they should fire for different presented objects (see http://precedings.nature.com/documents/5345/version/2).
(iii) A strong increase in the firing rate may show a different process
(iv)If spike directivity points randomly in space then the strong increase in firing rate display uncertainty, an ongoing “searching” process to associate different presented features
(v)The efficient coding of a particular feature associated with Jennifer Aniston needs to provide a consistent preferred spike directivity. This outcome is determined by a consistent intracellular location of particular "memories" and is revealed using spike directivity or imaging the spike. see http://precedings.nature.com/documents/5345/version/2 or http://neuroelectrodynamics.blogspot.com/p/spike-directivity.html
6. In your image control experiments, where the subject focused on one of two images on a computer screen to enhance its visibility (a target vs. a distractor image) by “thinking” about it (the target image), the subjects were able to control and modulate the activity of the concept units selective to specific images. “Thinking” in this case might simply imply invoking some images from memory of the target concept (e.g. Jennifer Aniston) and that might also imply the “internal assignment” of meaning to the target concept cells. (This is a tenuous argument. Wish we could also say that the concept cell activates the “memories,” thereby providing linkage both ways.)
Itzhak Fried: Read the Science 2008 paper by Gelbard -Sagiv, Mukamel, Harel, Malach and Fried where you will see how such cell (firing selectively to the actual sight of a 10 sec video of the Simpson's, as one example ) is reactivated just before (1.2 sec) the patient reports the recollection.
Christof: The cognitive or neuronal processes underlying the voluntary control seen in these fading experiments are unclear. Personally, I think they are closer to object based attention than to memory but this remains to be proven.
Dorian: I was particularly interested by the Simpson example, even stopped the presentation to show that this specific cell has fired for different other images with low firing rate. If during the increase in firing rate the computed spike directivities show outcome then this case can be a typical example where the neuron is “searching” for a solution. Since other different neurons may activate the recollection the re-searching process can be triggered. I feel that this neuron will not provide stable high firing rate longer time in this case (over one week of repeated Simpson presentation). If spike directivity is less random then indeed this particular neuron can embed some features associated to Simpson. Information is “read” or “written” in this cell during these spikes and the cell may contribute to form the Simpson abstraction( however not alone!).
7. Here’s an interesting conclusion from Waydo, Kraskov, Quiroga, Fried and Koch (The Journal of Neuroscience, 2006):
“Instead, it would imply that rather than a single neuron responding to dozens of stimuli out of a universe of tens of thousands, such a neuron might respond to only one or a few stimuli out of perhaps hundreds currently being tracked by this memory system, still with millions of neurons being activated by a typical stimulus. These results are consistent with Barlow’s (1972) claim that “at the upper levels of the hierarchy, a relatively small proportion [of neurons] are active, and each of these says a lot when it is active,” and his further speculation that the “aim of information processing in higher sensory centers is to represent the input as completely as possible by activity in as few neurons as possible” (Barlow, 1972).”
Itzhak Fried: When I proposed the term "concept cells" for the unique group of cells we found in hippocampus and neighbouring MTL structures it was with the intention of provoking such diuscussions, but using the nomenclature we should not be carried away by the hype of the terminology and lose sight of the data.(I do agree with Freeman's cautionary note re "meaning"). Do not forget that these cells are at the heart of the declarative memory system of MTL and thus signify the transformation of percepts into what can be later consciously recollected. The intriguing question is how these cells are formed and change. We know patients form these cells to the experimenters over a day or so. We are currently completing a study which will provide some relevant data.
Christof: Yes, Horace Barlow's 1972 paper was very forward looking and deserves to be widely read and cited. The efficiency in processing information is the main goal of the brain, therefore the process of object recognition is optimized in these cells that respond which a decrease in firing rate and a “specialization” of involved neurons that carries specific features
Quian Quiroga
Asim, thanks for triggering this interesting discussion.Yes, I do believe these cells encode meaning. We say this explicitly in a TiCS paper (at the end of the section before the Conclusion).
From Walter Freeman
Your paraphrase is ambiguous. "Concept cells" certainly have meaning for observers, but do they express and transmit meaning within the brain of the subject to other parts of the brain? In other words, how in a small fraction of a second does the output of the "concept cell" capture and control attention and the neural machinery leading to Sherrington's "final common path"? I conceive the "concept cell" as one of ~10^5 neurons forming a Hebbian assembly, which provides the key to a global attractor and the energy needed to trigger a phase transition. In this view the meaning is expressed by the attractor involving ~10^9 neurons. The spikes of the sampled "concept cell" (in concert with ~10^5 - 1 other cells) are an essential sign, neural correlate, and agency mediating the construction of meaning from the memory (synaptic matrix) selected by a stimulus
Dorian: The efficiency of information processing seems to be the main reason of changes in the dynamics of firing. The T-maze learning shows a process of optimization. The firing rate is reduced when a certain semantics is acquired in single cells. Here, in these recordings it seem to be a similar process. If the objects are presented several times the increase in “specialization” occurs http://neuroelectrodynamics.blogspot.com/p/spike-directivity.html).I agree with Dr Freeman. The temporal coding is ambiguous (see both counterexamples) the meaning seems to be a result of electrical inference (not of temporal patterns) http://precedings.nature.com/documents/5345/version/2) .The Horace Barlow's 1972 paper was an inspiration to develop the new computational model – NeuroElectroDynamics (NED). This computational model shows that information is integrated across different scales in the brain using electrical activity (not temporal patterns) in order to generate the “the final common path” -see a small network of four neurons http://precedings.nature.com/documents/5345/version/2 . Many scientists have previously envisioned and described different (non-Turing) forms of computations. Computing by physical interaction in neurons ( in the brain) generates a powerful (non-Turing) model of computation. I’m always interested to analyze good recordings and really delighted that with these new techniques the mystery of neural code can be solved http://neuroelectrodynamics.blogspot.com/p/cracking-neural-code.html “Cracking” the neural code was not the main goal . The result occurred in response to other different questions- Why artificial intelligence cannot move beyond capabilities of a two-and-a-half year old child? Why brain computations are so powerful? We found that several controversies in the field were generated by keeping alive the temporal coding paradigm. I value the contribution of all scientists that have worked in this field; they kept our interest focused on fundamental issues and our success in understanding how the brain computes (see NED) reflects their long-standing effort in this area.
"We have seen a little further by standing on the shoulders of Giants .... not because our sight is superior or because we are taller than they, but because they raise us up, and by their great stature add to ours”Isaac Newton
Therefore, I'm actively interacting to clarify several issues regarding temporal coding.
neuroelectrodynamics.blogspot.com/
http://neuronline.sfn.org/SFN/SFN/Home/Default.aspx (requires membership)
www.linkedin.com/groups/Computational-Neuroscience-1376707 (requires membership)
Therefore, I'm actively interacting to clarify several issues regarding temporal coding.
neuroelectrodynamics.blogspot.com/
http://neuronline.sfn.org/SFN/SFN/Home/Default.aspx (requires membership)
www.linkedin.com/groups/Computational-Neuroscience-1376707 (requires membership)
Thu, Aug 11, 2011 at 1:24 PM: Asim ROY: Please don’t post on this list (THIS IS THE BROAD LIST). You should have asked me to add you to the other list that is discussing this actively. Again, please don’t post to this list.
Thu, Aug 11, 2011 at 2:08 PM:Dorian AUR: Sorry for misunderstanding, I didn't know about the other list. I would really appreciate if you could add me there. Thank you,On Fri, Aug 12, 2011 at 12:06 AM, Asim Roy added me on the short list and my response on Aug 14
Asim Roy: I understand your issues better now than when you presented your poster at Moran Cerf’s talk at IJCNN. Here’s my read on what you are saying and the experiments of Fried, Koch, Cerf, Quian Quiroga and others.
(YOU MAY SEE, NO ISSUES REGARDING MORAN’S TALK, HAPPY TO UNDERSTAND THAT’S a SCIENTIFIC DEBATE)
Dorian Aur: First, I really appreciate your decision to add me on this list. Well, it is hard to reach across several disciplines and explain several issues in interpretation generated by temporal coding techniques. An introduction was written (Neuroelectrodynamics- NED) however additional explanations are still needed. A short article has just been published (From Neuroelectrodynamics to Thinking Machines http://www.springerlink.com/openurl.asp?genre=article&id=doi:10.1007/s12559-011-9106-3 ) that summarizes some aspects including the relationship between NED and connectionism.
Asim Roy:1. You are claiming that the firing rate or interspike interval of a single neuron (temporal coding) does not carry much information or meaning. However, if you look at the Moran Cerf experiments, they adjusted the images on the screen based on these firing rates. So there is external validation that the firing rate of a single neuron has information and meaning. One can claim that there are better indicators or measures, but the measure they used worked quite well.
Dorian Aur:Indeed, there is little information and the real “meaning” cannot be extracted from firing rate or ISI analysis. The result of temporal coding analysis is ambiguous. I hope that everyone can well understand both counterexamples.
The main issue is generated by the main hypothesis that all action potentials are similar (digital uniformity). From a false hypothesis (stereotypical spike) one can reach false conclusions, false interpretations following a correct statistical analysis of data. We have shown in Ch 4 in a simulation that information can be stored in a spatial distribution of electric charges and then moved in Ch5 to the real brain structure. The appearance of rhythmic "spikes" and temporal patterns carry almost no information regarding the “coded” feature. More useful information is within the generated spikes http://dx.doi.org/10.1038/npre.2010.5345.2.The temporal data analysis has generated so many contradictory and confusing results (see firing rate, ISI -Ch1 in Neuroelectrodynamics e.g. grandmother cells) .
The main issue is generated by the main hypothesis that all action potentials are similar (digital uniformity). From a false hypothesis (stereotypical spike) one can reach false conclusions, false interpretations following a correct statistical analysis of data. We have shown in Ch 4 in a simulation that information can be stored in a spatial distribution of electric charges and then moved in Ch5 to the real brain structure. The appearance of rhythmic "spikes" and temporal patterns carry almost no information regarding the “coded” feature. More useful information is within the generated spikes http://dx.doi.org/10.1038/npre.2010.5345.2.The temporal data analysis has generated so many contradictory and confusing results (see firing rate, ISI -Ch1 in Neuroelectrodynamics e.g. grandmother cells) .
Asim Roy:2. As far as I understand, none of the experiments in Dr. Fried’s lab involved learning. For example, Jennifer Aniston or Mother Teresa was not a concept taught to the patients during these experiments. Those were already existing concepts in these patients. So your analysis of firing rates during learning and after learning (T-maze learning in rats??) does not quite apply when you argue against their claims. They are simply looking at firing rates post-learning.
Dorian Aur: Dr Fried had an excellent idea, the experiments, image presentation, excellent recordings. However, there are unavoidable issues. The experiment requires a repetitive presentation of images therefore, a certain plasticity occurs during several days of repetitive presentation. The phenomenon is recognized "The intriguing question is how these cells are formed and change" therefore they all have observed the CHANGE.
They have observed this change in the T-maze at MIT (reorganization....) they were interested to analyze the change, however they didn’t relate the change with what happens within neurons.
The T-maze counterexample shows a more general principle and provides in fact the answer regarding this issue.
(i) Almost all phenomena in the brain involve a reorganization of neuronal activity. During repetitive events (image presentation, T-maze task) the activity within neurons is reorganized and leads to a "stable" state in the "post-learning" phase.
(ii)The reorganization (learning) can be seen as a transitory phase (less than one week in case of the T-maze) 4-7 days in some cells (1-2 days in case of re-training ).
(iii)The reorganization decreases the firing rate, increases the efficiency of spikes since semantics are acquired .
(iv)The post-learning (post-reorganization) shows a "sparse" activity and “order” in spike directivity (for left and right turn) which is directly related to semantics.
(v) The real meaning (semantics) does not appear in the firing rate or ISI see the counterexample (about the same firing rate for the left and the right turn http://neuroelectrodynamics.blogspot.com/p/spike-directivity.html
(vi)In case of repetitive task events (T-maze, UCLA experiment) the principle of efficient information processing determines a reorganization of information processing within neurons.
(vii)This example where spike directivity changes and remains stable in the post learning phase shows that during training reorganization occurs at a smaller scale inside the cell
They have observed this change in the T-maze at MIT (reorganization....) they were interested to analyze the change, however they didn’t relate the change with what happens within neurons.
The T-maze counterexample shows a more general principle and provides in fact the answer regarding this issue.
(i) Almost all phenomena in the brain involve a reorganization of neuronal activity. During repetitive events (image presentation, T-maze task) the activity within neurons is reorganized and leads to a "stable" state in the "post-learning" phase.
(ii)The reorganization (learning) can be seen as a transitory phase (less than one week in case of the T-maze) 4-7 days in some cells (1-2 days in case of re-training ).
(iii)The reorganization decreases the firing rate, increases the efficiency of spikes since semantics are acquired .
(iv)The post-learning (post-reorganization) shows a "sparse" activity and “order” in spike directivity (for left and right turn) which is directly related to semantics.
(v) The real meaning (semantics) does not appear in the firing rate or ISI see the counterexample (about the same firing rate for the left and the right turn http://neuroelectrodynamics.blogspot.com/p/spike-directivity.html
(vi)In case of repetitive task events (T-maze, UCLA experiment) the principle of efficient information processing determines a reorganization of information processing within neurons.
(vii)This example where spike directivity changes and remains stable in the post learning phase shows that during training reorganization occurs at a smaller scale inside the cell
Therefore, both, the reorganization (learning) and the "post-learning" (stable) phase were analyzed in the T-maze experiment. In fact the picture in http://neuroelectrodynamics.blogspot.com/p/spike-directivity.html is from the post-learning phase. Before the task is learnt, spike directivity is random. It is random in the post-learning phase, between gate opens and tone or between beginning of turn and goal (see, Neuroelectrodynamics and D Aur, and M Jog, Reading the Neural Code: What do Spikes Mean for Behavior?. Available from Nature Precedings <http://dx.doi.org/10.1038/npre.2007.61.1, 2007 ).
(a) The repetitive presentation of images generates the same behavior in MTL neurons. assuming that Jennifer Aniston or the spider were previously learnt.
(b)The “stable” directivity is visible in the post-learning phase (see the spider and Jennifer Aniston) ---
(c) About the same meaningless firing rate is recorded for the spider and Jennifer Aniston (see Aur D, Where is the ‘Jennifer Aniston neuron’? , available from Nature Precedings, fig 6 http://dx.doi.org/10.1038/npre.2010.5345.2). Sorry they didn’t include Mother Teresa.in this experiment
This simple counterexample shows that one cannot trust results generated by firing rate regarding semantics.
(a) The repetitive presentation of images generates the same behavior in MTL neurons. assuming that Jennifer Aniston or the spider were previously learnt.
(b)The “stable” directivity is visible in the post-learning phase (see the spider and Jennifer Aniston) ---
(c) About the same meaningless firing rate is recorded for the spider and Jennifer Aniston (see Aur D, Where is the ‘Jennifer Aniston neuron’? , available from Nature Precedings, fig 6 http://dx.doi.org/10.1038/npre.2010.5345.2). Sorry they didn’t include Mother Teresa.in this experiment
This simple counterexample shows that one cannot trust results generated by firing rate regarding semantics.
Asim Roy:3. Your first counter example (rats and T-maze): Could it be that the neuron was not a place cell, just some other cell??
Not sure if the name (concept, place, expert, mirror) tells anything. The meaning is just for us ---(excellent explanation from Dr. Freeman). The term "expert" neuron comes from Ann Graybiel, (MIT group) - may have a certain validity. In fact it can describe the activity of all types of cells that share our "meaning" when they fire. Therefore from a computational perspective the "concept" cells do not bring anything special, they have to be included in this category of "expert" neurons. They fire for events that WE consider meaningful. The "other" cell can be recruited and turned into an "expert" cell in 4-7 days or 1-2 days in case of re-training (the “same” cell). The time-line can be different in MTL. The grandiose terms (place cell, concept cell, mirror neurons, Jennifer Aniston cell ...) tend to perpetuate the gimmick of temporal coding theory, the myth of temporal coding. They do not reflect a deep understanding of the phenomena.
Not sure if the name (concept, place, expert, mirror) tells anything. The meaning is just for us ---(excellent explanation from Dr. Freeman). The term "expert" neuron comes from Ann Graybiel, (MIT group) - may have a certain validity. In fact it can describe the activity of all types of cells that share our "meaning" when they fire. Therefore from a computational perspective the "concept" cells do not bring anything special, they have to be included in this category of "expert" neurons. They fire for events that WE consider meaningful. The "other" cell can be recruited and turned into an "expert" cell in 4-7 days or 1-2 days in case of re-training (the “same” cell). The time-line can be different in MTL. The grandiose terms (place cell, concept cell, mirror neurons, Jennifer Aniston cell ...) tend to perpetuate the gimmick of temporal coding theory, the myth of temporal coding. They do not reflect a deep understanding of the phenomena.
Asim Roy:4. The second counter example (a cell that responds to both a spider and Jennifer Aniston): Was it one of the concept cells reported in these studies as a Jennifer Aniston cell or a spider cell? Are you claiming that they have misrepresented the data? Or perhaps this was another cell, not a concept cell, that responded to both concepts and you are claiming that other measures, like spike directivity, would have done a better job in distinguishing the concepts?
Dorian Aur: That's a very good question. My daughter always asks if Santa Clause is real. Can I disappoint her? We like to believe in “place” neurons, “mirror” neurons “Jennifer” neuron ..... The question is: Are they real? Would you accept that "concept" neurons do not really exist?
To generate the Jennifer Aniston "concept" information is inferred from many "expert" neurons that embed information related to Jennifer Aniston. In fact if the cell fires to much to "Jennifer Aniston" it may show uncertainty, not a stable outcome where "a characteristic part" of Jennifer Aniston is “read” or “written” during the generated spikes. In this case they have clearly misrepresented the data and “Jennifer Aniston” neuron can be another “amazing” result generated by the ambiguous analysis of temporal patterns.
If this neuron is an “expert” neuron that is related to "Jennifer " concept the analysis of spike directivity is expected to show a more stable outcome (see http://neuroelectrodynamics.blogspot.com/p/spike-directivity.html). However, the so called Jennifer neuron will always respond to other different images (e.g spider).
I wrote a paper regarding the myth of “Jennifer Aniston” neuron ---"Where is the Jennifer Aniston neuron?" http://precedings.nature.com/documents/5345/version/2 . A corrected version should be soon published.
Asim Roy: 5. You claim “a too strong increase of firing rate may show high uncertainty.” However, all of the experiments being done by Moran Cerf seem to take for granted that a higher firing rate implies a certain image or controls the movement of a ball on the screen and so on, in a very deterministic way. And they are successful in their experiments. So the obvious question is, where is the uncertainty in their interpretation?With temporal coding models, we all took for granted many things ( a very long list!). The increase of firing rate requires energy to generate APs. Therefore, the repetition of the same events (image presentation, T-maze) determines a reorganization needed to generate an efficient optimal response within few spikes. Indeed, a complex task requires an increase of computational power and many neurons are required to fire simultaneously to generate a solution in the first phase (see the explanation in NED). However, in case of repetitive events, a general a stable (well known) response should not be generated with a strong increase in the firing rate. I completely agree with the question “where is the uncertainty in their interpretation?”
In addition, the claim that reorganization (learning) does not occur in a repetitive task is completely untrue. I would say that reorganization occurs even faster if the concept was learnt before (see the T-maze task example).
To generate the Jennifer Aniston "concept" information is inferred from many "expert" neurons that embed information related to Jennifer Aniston. In fact if the cell fires to much to "Jennifer Aniston" it may show uncertainty, not a stable outcome where "a characteristic part" of Jennifer Aniston is “read” or “written” during the generated spikes. In this case they have clearly misrepresented the data and “Jennifer Aniston” neuron can be another “amazing” result generated by the ambiguous analysis of temporal patterns.
If this neuron is an “expert” neuron that is related to "Jennifer " concept the analysis of spike directivity is expected to show a more stable outcome (see http://neuroelectrodynamics.blogspot.com/p/spike-directivity.html). However, the so called Jennifer neuron will always respond to other different images (e.g spider).
I wrote a paper regarding the myth of “Jennifer Aniston” neuron ---"Where is the Jennifer Aniston neuron?" http://precedings.nature.com/documents/5345/version/2 . A corrected version should be soon published.
Asim Roy: 5. You claim “a too strong increase of firing rate may show high uncertainty.” However, all of the experiments being done by Moran Cerf seem to take for granted that a higher firing rate implies a certain image or controls the movement of a ball on the screen and so on, in a very deterministic way. And they are successful in their experiments. So the obvious question is, where is the uncertainty in their interpretation?With temporal coding models, we all took for granted many things ( a very long list!). The increase of firing rate requires energy to generate APs. Therefore, the repetition of the same events (image presentation, T-maze) determines a reorganization needed to generate an efficient optimal response within few spikes. Indeed, a complex task requires an increase of computational power and many neurons are required to fire simultaneously to generate a solution in the first phase (see the explanation in NED). However, in case of repetitive events, a general a stable (well known) response should not be generated with a strong increase in the firing rate. I completely agree with the question “where is the uncertainty in their interpretation?”
In addition, the claim that reorganization (learning) does not occur in a repetitive task is completely untrue. I would say that reorganization occurs even faster if the concept was learnt before (see the T-maze task example).
Asim Roy:3. You say: “The firing rate is reduced when a certain semantics is acquired in single cells.” Does that mean that you are really not questioning the idea of single cells having meaning? It appears to me that you are arguing that they have a wrong way of finding such concept cells and that you have a better way of finding them. And you could be right. But their measure works too. And they have demonstrated that in these experiments, unless you claim that have misrepresented the data and the claims are false.
The collected data in Dr Fried lab is excellent. However, unfortunately it seems that based on firing rate analysis they have misinterpreted the data. Indeed, the analysis of spikes can show the meaning, the T-maze experiment is the proof. The finer scale of information processing within the cells can provide this information in a more reliable fashion, it shows the real meaning of recorded spikes, the “neural code”. The real meaning does not explicitly appear in the firing rate (see both counterexamples about the same firing rate, completely different meaning! That's the entire philosophy of counterexamples.
The T-maze procedural task offers a more general perspective since it relates the analyzed behavior to habit formation and reward acquisition. These results can even explain why temporal coding theory “founded on nothing” was kept alive so long time without any significant results (new theories, science breakthroughs). We need to understand that during the last 50 years in neuroscience getting the “reward” was conditioned by an attachment to spike timing dogma (STD). On a different forum I’ve asked what are the major theories (science breakthroughs) that have resulted from firing rate or ISI analyzes in the last twenty years . They pointed to “reinforcement learning” even to Bayesian models.... they do not know about "Jennifer Aniston" neuron even though was highly advertised
Indeed, many papers have mapped an already existent model (reinforcement learning model; probability, Bayesian model ) adapted it to fit temporal patterns. Given the complexity of temporal patterns in the brain anyone can find neurons that appear to follow a specific model ( reinforcement ; probabilistic, Bayes, linear-nonlinear with just three parameters .....) and imagine explanations that provide OUR meaning. Adding together a good mathematical theory to temporal coding paradigms does not make the spike timing theory a better model. All experiments show a more powerful model of computation in neurons and in the brain. The current data analysis of temporal patterns in neuroscience is an insult to the capabilities of actual neurons and brain computations, the EXISTING PARADIGM WHICH IS BASED ON TEMPORAL CODING is completely MISLEADING. I feel that the computation in the brain deserves a specific model, NED is just an introduction. The entire connectionist theory can be seen as a particular model of interaction, I'm planning to write more regarding this relationship. An entire infrastructure needs to be built to allow a different approach, even new journals ... I'm looking for collaborators that can understand that brain computations can not be described by spike timing dogma and hope that together we can change this outcome.
On Sun, Aug 14, 2011 at 7:40 PM If I understand correctly, you are questioning all of neuroscience that believes that spike timing has information (your term: “spike timing dogma”), not just the results from Dr. Fried’s lab. Here’s some of what you said:
“These results can even explain why temporal coding theory “founded on nothing” (term borrowed from from http://mathematicallinguistics.blogspot.com/2011/07/neuroelectrodynamics.html) was kept alive so long time without any significant results (new theories, science breakthroughs). We need to understand that during the last 50 years in neuroscience getting the “reward” was conditioned by an attachment to spike timing dogma (STD). On a different forum I’ve asked what are the major theories (science breakthroughs) that have resulted from firing rate or ISI analyzes in the last twenty years . They pointed to “reinforcement learning” even to Bayesian models.... they do not know about "Jennifer Aniston" neuron even though was highly advertised.”
Here’s a summary of the experiments and results reported in Cerf, Thiruvengadam, Mormann, Kraskov, Quian-Quiroga, Koch, & Fried (2010):
“In their experiments, twelve epilepsy patients played a game where they controlled the display of two superimposed images. The controlling was done through the activity of four MTL neurons. Before the actual experiment, the researchers identified four different neurons that responded selectively to four different images. In these experiments, one of the four images was randomly designated as the target image. Each trial started with a short display of a random target image followed by an overlaid hybrid image consisting of the target and one of the other three images (the distractor image). The subject was then told to enhance the target image by focusing his/her thoughts on it; as per Cerf et al. (2010), the patients were instructed to ‘‘continuously think of the concept represented by that image.” The initial visibility of both the images was 50% and the visibility of an image was increased or decreased every 100 ms based on the firing rates of the four MTL neurons that were monitored. Spike counts were actually used to measure firing rates and to control the visibility of the two images on the screen. In general, if the firing rate of one neuron was higher compared to the other, more visible was the image associated with that neuron and less visible the other image. The trial was terminated when either one of the two images, the target or the distractor image, was fully visible or after a fixed time limit of 10 seconds. The subjects successfully reached the target in 596 out of 864 trials (69.0%; 202 failures and 66 timeouts).”
If your arguments are correct, that spike timing or firing rates have no information, then they couldn’t have got these results. It’s as simple as that. Here are samples of your arguments that temporal coding has no information:
1. “Indeed, there is little information and the real “meaning” cannot be extracted from firing rate or ISI analysis. The result of temporal coding analysis is ambiguous.”
2. “The real meaning (semantics) does not appear in the firing rate or ISI see the counterexample (about the same firing rate for the left and the right turn”
3. “This simple counterexample shows that one cannot trust results generated by firing rate regarding semantics.”
4. “The real meaning does not explicitly appear in the firing rate (see both counterexamples about the same firing rate, completely different meaning! That's the entire philosophy of counterexamples.”
5. “The collected data in Dr Fried lab is excellent. However, unfortunately it seems that based on firing rate analysis they have misinterpreted the data. Indeed, the analysis of spikes can show the meaning, the T-maze experiment is the proof.”
2. “The real meaning (semantics) does not appear in the firing rate or ISI see the counterexample (about the same firing rate for the left and the right turn”
3. “This simple counterexample shows that one cannot trust results generated by firing rate regarding semantics.”
4. “The real meaning does not explicitly appear in the firing rate (see both counterexamples about the same firing rate, completely different meaning! That's the entire philosophy of counterexamples.”
5. “The collected data in Dr Fried lab is excellent. However, unfortunately it seems that based on firing rate analysis they have misinterpreted the data. Indeed, the analysis of spikes can show the meaning, the T-maze experiment is the proof.”
Asim Roy :If firing rates had no meaning, then Moran Cerf experiments wouldn’t have worked. It as simple as that. So you have to explain why Moran Cerf experiments worked.
AUG 15, 2011 AT 2:10 PM THE NEW TOPIC OF DISCUSSION TO MORAN CERF EXPERIMENTS
Tue, Aug 16, 2011 at 11:55 AM Dorian Aur: You’re right, no response yet to Moran Cerf experiment. However, I would like first to know if everyone on this list agrees with the two fundamental aspects 1. the false hypothesis of a digital uniform (stereotype) action potential 2. that the counter example model can be used in neuroscience. Please, confirm or tell me if there are any issues, then we will move to Moran Cerf experiment
I RECEIVED MESSAGES THAT THEY CANNOT SEE AND UNDERSTAND THE IMAGES ON THE BLOGSPOT http://neuroelectrodynamics.blogspot.com/
On Mon, Aug 15, 2011 at 2:10 PM, Dorian Aur: Therefore anyone should be able to see all pages on http://neuroelectrodynamics.blogspot.com/
* Home
* Temporal Coding - a Realistic Model of Neural Act...
* Spike Directivity
* Cracking the Neural Code
* Home
* Temporal Coding - a Realistic Model of Neural Act...
* Spike Directivity
* Cracking the Neural Code
(i) About the same firing rate is recorded in the neuron (8Hz) during both presentations (spider, Jennifer Aniston)
(ii) Consistently different parts of recorded neuron are activated when the spider or Jennifer Aniston images are presented .(see Spike Directivity)
The neuron responds to the presented event and tells that a certain information regarding this event is “read” or “written” during the spike.
Based on spike directivity one can predict that Jennifer Aniston was presented and not the spider. Therefore, the firing rate does not tell WHAT information was “read” or “written”. In fact the firing rate doesn’t tell the meaning (semantics)
(iii) This is a simple counter example which shows that information regarding meaning (semantics) is hidden and is not available in the firing rate.
(iv)The counter example is obtained from their experiment, strongly refutes the temporal coding hypothesis.
(v) Therefore, any previous experiment that has proven that firing rate is a good measure to discriminate the meaning (semantics) or claims the existence of "temporal coding" has to be reanalyzed and reinterpreted. That’s the idea of counterexample presentation
(ii) Consistently different parts of recorded neuron are activated when the spider or Jennifer Aniston images are presented .(see Spike Directivity)
The neuron responds to the presented event and tells that a certain information regarding this event is “read” or “written” during the spike.
Based on spike directivity one can predict that Jennifer Aniston was presented and not the spider. Therefore, the firing rate does not tell WHAT information was “read” or “written”. In fact the firing rate doesn’t tell the meaning (semantics)
(iii) This is a simple counter example which shows that information regarding meaning (semantics) is hidden and is not available in the firing rate.
(iv)The counter example is obtained from their experiment, strongly refutes the temporal coding hypothesis.
(v) Therefore, any previous experiment that has proven that firing rate is a good measure to discriminate the meaning (semantics) or claims the existence of "temporal coding" has to be reanalyzed and reinterpreted. That’s the idea of counterexample presentation
Aug 15, 2011 at 7:52 PM Asim Roy:You need to explain why the Moran Cerf experiments on image control based on firing rates of particular neurons worked. If your argument/theory is right, that temporal coding has no meaning, then those experiments should have failed. You need to explain why they worked. Explaining this kind of experimental result is crucial to your argument against temporal coding (“spike timing dogma”).
Tue, Aug 16, 2011 at 9:42 AM Dorian Aur : You've asked excellent questions, I do not avoid a response. First, I would like to be sure that everyone on this list can see these images on http://neuroelectrodynamics.blogspot.com/ . Today, I already have the confirmation . Second, that everyone understands well the experimental results. I'm sure that understanding this outcome provides a response to many other questions . ( AND I WAS RIGHT!)
There are two fundamental aspects that one needs to understand:
1. The false hypothesis of a digital uniform (stereotype) action potential . If anyone on this list feels that action potentials generated by a neuron are uniform (stereotype) please explain
2.The counter example model can be used in neuroscience. If anyone on this list feels that the counter example model cannot be used in neuroscience, please explain
There are two fundamental aspects that one needs to understand:
1. The false hypothesis of a digital uniform (stereotype) action potential . If anyone on this list feels that action potentials generated by a neuron are uniform (stereotype) please explain
2.The counter example model can be used in neuroscience. If anyone on this list feels that the counter example model cannot be used in neuroscience, please explain
Tue, Aug 16, 2011 at 11:12 AM Asim Roy:The basic question remains unanswered: Why did the Moran Cerf experiments work if you think temporal coding has no information? Cerf used firing rates of concept cells in MTL for image control and the experiments were very successful. The Cerf experiments are a great validation that firing rates of certain cells have meaning and that significant information is associated with them. It certainly disproves your theory that firing rates and temporal coding have no meaning.
Tue, Aug 16, 2011 at 11:42 PM Asim Roy: Still no answer to the question why Moran Cerf experiments worked if your theory is correct. Those experiments were solely based on firing rates and they shouldn’t have worked, as per your argument. You say:
1. Dorian Aur “Therefore there is an issue in the Moran Cerf experiment, they basically cannot prove that semantics is in the firing rate.”
2. Dorian Aur“All these experiments have something in common . They do not prove that the code is in the firing rate at all(or that information regarding semantics can be decoded using the firing rate from single neurons) . What is the phenomenon that they prove?”
They have repeated their experiment on a number of patients and it worked and that proves that firing rates have information. That’s science, when you can repeatedly demonstrate a certain outcome under similar conditions. Moran Cerf experiments definitely disproves your theory that there is no information in temporal codes.
On Tue, Aug 16, 2011 at 11:55 AM, Dorian Aur:You right, no response yet to Moran Cerf experiment. However, I would like first to know if everyone on this list agrees with the two fundamental aspects 1. the false hypothesis of a digital uniform (stereotype) action potential 2. that the counter example model can be used in neuroscience. Please, confirm or tell me if there are any issues, then we will move to Moran Cerf experiment
NO RESPONSE, THE LIST WAS QUIET THEREFORE MOVED FORWARD, SMALL STEPS TO REVEAL THE TRUTH ABOUT MORAN CERF EXPERIMENTS BY PROVIDING TIPS TO THE ENTIRE LIST
Tue, Aug 16, 2011 at 4:55 PM:Dorian Aur: I can see that everyone on this list agrees with both issues. The preferred spike directivity orientation in both experiments suggests the information that characterizes the presented object is stored in a topographic manner in neurons, a certain part of the neuron embeds the “memory” (semantics). In addition, this information does not explicitly appear in the temporal code (firing rate, ISI) . Therefore there is an issue in the Moran Cerf experiment, they basically cannot prove that semantics is in the firing rate.PROVIDING TIPS SO ANYONE COULD UNDERSTAND THE ISSUE
Tue, Aug 16, 2011 at 4:55 PM Dorian Aur: In order to clarify this issue I added other examples that use temporal patterns and pretend that with their interface they can "control" or "read" brain information.1. patients that move the cursor on the screen (Donoghue group also has claimed to have “cracked the neural code”)
Neuromotor prostheses (NMPs) aim to replace or restore lost motor functions in paralysed humans by routeing movement-related signals from the brain, around damaged parts of the nervous system, to external effectors. To translate preclinical results from intact animals to a clinically useful NMP, movement signals must persist in cortex after spinal cord injury and be engaged by movement intent when sensory inputs and limb movement are long absent. Furthermore, NMPs would require that intention-driven neuronal activity be converted into a control signal that enables useful tasks. Here we show initial results for a tetraplegic human (MN) using a pilot NMP. Neuronal ensemble activity recorded through a 96-microelectrode array implanted in primary motor cortex demonstrated that intended hand motion modulates cortical spiking patterns three years after spinal cord injury. Decoders were created, providing a 'neural cursor' with which MN opened simulated e-mail and operated devices such as a television, even while conversing. Furthermore, MN used neural control to open and close a prosthetic hand, and perform rudimentary actions with a multi-jointed robotic arm. These early results suggest that NMPs based upon intracortical neuronal ensemble spiking activity could provide a valuable new neurotechnology to restore independence for humans with paralysis.
(see Neuronal ensemble control of prosthetic devices by a human with tetraplegia
http://www.nature.com/nature/journal/v442/n7099/abs/nature04970.html)
2. Moran Cerf, On-line, voluntary control of human temporal lobe
n their experiments, twelve epilepsy patients played a game where they controlled the display of two superimposed images. The controlling was done through the activity of four MTL neurons. Before the actual experiment, the researchers identified four different neurons that responded selectively to four different images. In these experiments, one of the four images was randomly designated as the target image. Each trial started with a short display of a random target image followed by an overlaid hybrid image consisting of the target and one of the other three images (the distractor image). The subject was then told to enhance the target image by focusing his/her thoughts on it; as per Cerf et al. (2010), the patients were instructed to ‘‘continuously think of the concept represented by that image.” The initial visibility of both the images was 50% and the visibility of an image was increased or decreased every 100 ms based on the firing rates of the four MTL neurons that were monitored. Spike counts were actually used to measure firing rates and to control the visibility of the two images on the screen. In general, if the firing rate of one neuron was higher compared to the other, more visible was the image associated with that neuron and less visible the other image. The trial was terminated when either one of the two images, the target or the distractor image, was fully visible or after a fixed time limit of 10 seconds. The subjects successfully reached the target in 596 out of 864 trials (69.0%; 202 failures and 66 timeouts).”
3.Dr Tzyy-Ping Jung, San Diego that uses 16 EEG sensors to track cognitive states and dial
http://www.nature.com/nature/journal/v442/n7099/abs/nature04970.html)
2. Moran Cerf, On-line, voluntary control of human temporal lobe
n their experiments, twelve epilepsy patients played a game where they controlled the display of two superimposed images. The controlling was done through the activity of four MTL neurons. Before the actual experiment, the researchers identified four different neurons that responded selectively to four different images. In these experiments, one of the four images was randomly designated as the target image. Each trial started with a short display of a random target image followed by an overlaid hybrid image consisting of the target and one of the other three images (the distractor image). The subject was then told to enhance the target image by focusing his/her thoughts on it; as per Cerf et al. (2010), the patients were instructed to ‘‘continuously think of the concept represented by that image.” The initial visibility of both the images was 50% and the visibility of an image was increased or decreased every 100 ms based on the firing rates of the four MTL neurons that were monitored. Spike counts were actually used to measure firing rates and to control the visibility of the two images on the screen. In general, if the firing rate of one neuron was higher compared to the other, more visible was the image associated with that neuron and less visible the other image. The trial was terminated when either one of the two images, the target or the distractor image, was fully visible or after a fixed time limit of 10 seconds. The subjects successfully reached the target in 596 out of 864 trials (69.0%; 202 failures and 66 timeouts).”
3.Dr Tzyy-Ping Jung, San Diego that uses 16 EEG sensors to track cognitive states and dial
http://www.youtube.com/watch?v=P0HT3AC6BSIAll these experiments have something in common . They do not prove that the code is in the firing rate at all(or that information regarding semantics can be decoded using the firing rate from single neurons) . What is the phenomenon that they prove?
NO RESPONSE, A VERY QUIET LIST
On Wed, Aug 17, 2011 at 12:06 AM, Asim Roy Sorry, some got dropped from the earlier list. So I am reposting. Still no answer to the question why Moran Cerf experiments worked if your theory is correct. Those experiments were solely based on firing rates and they shouldn’t have worked, as per your argument. You say:
1. “Therefore there is an issue in the Moran Cerf experiment, they basically cannot prove that semantics is in the firing rate.”
2. “All these experiments have something in common . They do not prove that the code is in the firing rate at all(or that information regarding semantics can be decoded using the firing rate from single neurons) . What is the phenomenon that they prove?” They have repeated their experiment on a number of patients and it worked and that proves that firing rates have information. That’s science, when you can repeatedly demonstrate a certain outcome under similar conditions. Moran Cerf experiments definitely disproves your theory that there is no information in temporal codes.
1. “Therefore there is an issue in the Moran Cerf experiment, they basically cannot prove that semantics is in the firing rate.”
2. “All these experiments have something in common . They do not prove that the code is in the firing rate at all(or that information regarding semantics can be decoded using the firing rate from single neurons) . What is the phenomenon that they prove?” They have repeated their experiment on a number of patients and it worked and that proves that firing rates have information. That’s science, when you can repeatedly demonstrate a certain outcome under similar conditions. Moran Cerf experiments definitely disproves your theory that there is no information in temporal codes.
I REALIZED THAT ASIM TRIES TO MISINFORM (NO INFORMATION IN TEMPORAL CODES) AND REPLIED BACK
On Wed, Aug 17, 2011 12:29 PM, Dorian Aur: There is a subtle difference between our claims and it is excellent that you have asked.
Asim Roy:Moran Cerf experiments definitely disproves your theory that there is no information in temporal codes.
Dorian Aur: They do not prove that the code is in the firing rate at all or that information regarding semantics can be decoded using the firing rate from single neurons (there is only little information in the firing rate in single neurons). They show a completely different phenomenon that explains everything ---What is the phenomenon that they prove? One element is common in all three experiments. What's the specific phenomenon?
THAT’S ANOTHER KEY MOMENT, SILENCE ON THE ENTIRE LIST. I HOPED THAT SOMEONE ON THE LIST WILL SEE AND POINT TO THE FACT THAT MORAN CERF PAPER INVOLVES A COMPLETE DIFFERENT ASPECT.
Asher Evans had posted a message in support of Moran Cerf experiment and I’ve asked him the same question:
Dorian Aur Aug 17, 2011 at 3:58 PM Asher, What is common in all three experiments?
AUG 17, 2011 AT 5:17 PM OTHER MEMBERS ON THE LIST STARTED TO SEE THE ISSUE THAT “ACTION POTENTIALS HAVE CONSIDERABLE VARIATION AND THAT THIS VARIATION COULD CARRY INFORMATION”.THAT’S THE CRITICAL MOMENT WHEN THE KOCH GROUP REALIZED THAT SOMETHING IS WRONG and THE FEAR THAT THE TRICK REGARDING MORAN CERF EXPERIMENTS WILL BE REVEALED. UNTIL THIS POINT NONE OF THEM EXCEPT ASIM ROYHAS COMMENTED, HOWEVER THERE IS AN ISSUE THAT CAN BE REVEALED AND...
Wed, Aug 17, 2011 at 5:44 PM Christof Koch: That is right Asher. I can make two statement with 100% certainty, independent about any assumption of mind/brain or what the code is.
(1) The analysis in Quian-Quiroga, Reddy, Koch & Fried (J. Neurophysiol 1997) shows in an objective manner that the identity of visual stimuli that the patient is looking at can be predicted from the number of spikes fired by a handful of MTL cells (about 4 spikes per neuron, triggered between 300 and 600 msec after image onset in 7.8 units) far above chance.
(2) Moran's experiment shows that people can control a display by modulating the firing rate of four medial temporal lobe neurons using their thoughts (we're not sure what they are doing in the privacy of their heads; neither are they, of course). So, at the very least, the number of spikes discharged by specific neurons carries information and can causally influence events outside the brain (using feedback).
THEY REALIZED THAT THEIR ENTIRE PROPAGANDA USING MORAN CERF EXPERIMENT TO PROMOTE CONCEPT CELLS IDEA IS USELESS AND ASKED ASIM TO PRESS
17, 2011 at 8:26 PM Asim Roy: Unless you can explain, using your theory, why Moran Cerf experiments worked, it is a meaningless discussion.
THE NED TOPIC MADE NO SENSE AT THIS POINT, THE PROBLEM WAS MORAN CERF EXPERIMENT OTHERS FOUND CRISTOF KOCH COMMENTS TOO STRONG
Wed, Aug 17, 2011 at 11:14 PM Christof, are you sure (100% certainty) about the above phrase?
CRISTOF KOCH IGNORES THE QUESTION. PONTING TO THE REAL ISSUE OF MORAN CERF EXPERIMENT IN INTERPRETING THE CONCEPT CELL IDEA.THE BRAIN AREA FROM WHICH THE RECORDINGS ARE TAKEN IS INVOLVED IN CONTROLLING OR RESPONDING TO THE PRESENTED VARIABLE AND THE PATIENT RECEIVES FEEDBACK ABOUT BRAIN ACTIVITY
Asim Roy:Moran Cerf experiments definitely disproves your theory that there is no information in temporal codes.
Dorian Aur: They do not prove that the code is in the firing rate at all or that information regarding semantics can be decoded using the firing rate from single neurons (there is only little information in the firing rate in single neurons). They show a completely different phenomenon that explains everything ---What is the phenomenon that they prove? One element is common in all three experiments. What's the specific phenomenon?
THAT’S ANOTHER KEY MOMENT, SILENCE ON THE ENTIRE LIST. I HOPED THAT SOMEONE ON THE LIST WILL SEE AND POINT TO THE FACT THAT MORAN CERF PAPER INVOLVES A COMPLETE DIFFERENT ASPECT.
Asher Evans had posted a message in support of Moran Cerf experiment and I’ve asked him the same question:
Dorian Aur Aug 17, 2011 at 3:58 PM Asher, What is common in all three experiments?
AUG 17, 2011 AT 5:17 PM OTHER MEMBERS ON THE LIST STARTED TO SEE THE ISSUE THAT “ACTION POTENTIALS HAVE CONSIDERABLE VARIATION AND THAT THIS VARIATION COULD CARRY INFORMATION”.THAT’S THE CRITICAL MOMENT WHEN THE KOCH GROUP REALIZED THAT SOMETHING IS WRONG and THE FEAR THAT THE TRICK REGARDING MORAN CERF EXPERIMENTS WILL BE REVEALED. UNTIL THIS POINT NONE OF THEM EXCEPT ASIM ROYHAS COMMENTED, HOWEVER THERE IS AN ISSUE THAT CAN BE REVEALED AND...
Wed, Aug 17, 2011 at 5:44 PM Christof Koch: That is right Asher. I can make two statement with 100% certainty, independent about any assumption of mind/brain or what the code is.
(1) The analysis in Quian-Quiroga, Reddy, Koch & Fried (J. Neurophysiol 1997) shows in an objective manner that the identity of visual stimuli that the patient is looking at can be predicted from the number of spikes fired by a handful of MTL cells (about 4 spikes per neuron, triggered between 300 and 600 msec after image onset in 7.8 units) far above chance.
(2) Moran's experiment shows that people can control a display by modulating the firing rate of four medial temporal lobe neurons using their thoughts (we're not sure what they are doing in the privacy of their heads; neither are they, of course). So, at the very least, the number of spikes discharged by specific neurons carries information and can causally influence events outside the brain (using feedback).
THEY REALIZED THAT THEIR ENTIRE PROPAGANDA USING MORAN CERF EXPERIMENT TO PROMOTE CONCEPT CELLS IDEA IS USELESS AND ASKED ASIM TO PRESS
17, 2011 at 8:26 PM Asim Roy: Unless you can explain, using your theory, why Moran Cerf experiments worked, it is a meaningless discussion.
THE NED TOPIC MADE NO SENSE AT THIS POINT, THE PROBLEM WAS MORAN CERF EXPERIMENT OTHERS FOUND CRISTOF KOCH COMMENTS TOO STRONG
Wed, Aug 17, 2011 at 11:14 PM Christof, are you sure (100% certainty) about the above phrase?
CRISTOF KOCH IGNORES THE QUESTION. PONTING TO THE REAL ISSUE OF MORAN CERF EXPERIMENT IN INTERPRETING THE CONCEPT CELL IDEA.THE BRAIN AREA FROM WHICH THE RECORDINGS ARE TAKEN IS INVOLVED IN CONTROLLING OR RESPONDING TO THE PRESENTED VARIABLE AND THE PATIENT RECEIVES FEEDBACK ABOUT BRAIN ACTIVITY
Aug 18, 2011 at 10:16 AM Dorian Aur: I didn't want to reveal it right away since any of you should experience the joy of discovering. Therefore, today I was just planning to provide some tips to point to the trick. However, Dr Koch felt that soon enough one of you will understand the issue regarding Moran Cerf experiment and has decided himself to reveal it. In this case the explanation should start with (2) not with (1).
Therefore, Dr Koch knows well the trick : the patient can control the firing rate and modulate the image on the display using the feedback (it was revealed by many others before and is common all three above examples) . The paper does not demonstrate the concept cell idea!!!
Therefore, Dr Koch knows well the trick : the patient can control the firing rate and modulate the image on the display using the feedback (it was revealed by many others before and is common all three above examples) . The paper does not demonstrate the concept cell idea!!!
Therefore all my previous statements remain valid :
1. They do not prove that the "code" is in the firing rate at all - which is equivalent to- Moran Cerf experiments didn't crack the neural code using the firing rate (similar experiment did Donoghue group before and they didn't crack the code either).
2. or that information regarding semantics can be decoded using the firing rate from single neurons .
3. Since Moran Cerf experiment is revealed we found that in fact another different experiment Quian et al. in 1997 has predicted from the number of spikes fired by a handful of MTL cells the identity of visual stimuli,
A handful of MTL cells is different than the firing rate in a single neuron (see the Jennifer Aniston example) and the the "identity of visual stimuli" not sure exactly if this represents the concept cell idea that were presented as Moran Cerf results....
This unexpected outcome, (the papers are switched) shows that in fact was right before:
4. From a false hypothesis (stereotype spike ) following a correct algorithm (statistics) one can demonstrate that Santa Claus is real using brain recordings (complex, rich in temporal patterns).
1. They do not prove that the "code" is in the firing rate at all - which is equivalent to- Moran Cerf experiments didn't crack the neural code using the firing rate (similar experiment did Donoghue group before and they didn't crack the code either).
2. or that information regarding semantics can be decoded using the firing rate from single neurons .
3. Since Moran Cerf experiment is revealed we found that in fact another different experiment Quian et al. in 1997 has predicted from the number of spikes fired by a handful of MTL cells the identity of visual stimuli,
A handful of MTL cells is different than the firing rate in a single neuron (see the Jennifer Aniston example) and the the "identity of visual stimuli" not sure exactly if this represents the concept cell idea that were presented as Moran Cerf results....
This unexpected outcome, (the papers are switched) shows that in fact was right before:
4. From a false hypothesis (stereotype spike ) following a correct algorithm (statistics) one can demonstrate that Santa Claus is real using brain recordings (complex, rich in temporal patterns).
ASIM ROY DOESN’T SEEM TO UNDERSTAND THE BIG ISSUE.
Aug 18, 2011 at 10:56 AM Asim Roy:You said: “The paper does not demonstrate the concept cell idea!!!” They found those four concept cells, corresponding to the four different objects, prior to the actual experiment. Moran discussed that in his presentation at IJCNN and you were there. And it’s in their paper.
Thu, Aug 18, 2011 at 11:31 AM Dorian Aur: The Moran Cerf paper does not demonstrate the concept cell idea, it demonstrates a completely different thing. The "concept cell" is the grandmother idea, otherwise you'll call it "many concepts in the cell". I'm not sure if other scientists will agree.
Thu, Aug 18, 2011 at 11:38 AM Cristof Koch (only to Asim and me) Concept cells, at least as we define them, are very different from grandmother cells. Of course, you're free to define them in any way you feel like but that's not our def
SHORT TIME AFTER ASIM ON THE ENTIRE LIST:
Thu, Aug 18, 2011 at 12:00 PM Asim Roy: They are not claiming that concept cells are grandmother cells. They have never claimed that in any of their papers. And they have said that many times in these discussions and in the talk by Moran Cerf. As far as I understand, concept cells are part of a very sparse representation and they have been very successful in finding them in their experiments, because, by their estimate, 40% of the cells in MTL are estimated to be concept cells. And because the concept cells are part of a very sparse representation, they are very highly informative. They usually find one of them for each concept and use them in their experiments, such as the one by Moran Cerf to control images.
THU, AUG 18, 2011 AT 12:11 PM THE IDEA OF REGARDING CONCEPT CELL IS STRONGLY QUESTIONED BY OTHERS. THE KOCH GROUP HAVE TO ANSWER IF THEY “HAVE A CLEAR, SPECIFIC DEFINITION OF THE CONCEPT CELL IDEA OR GRANDMOTHER CELLS”. ASIM ROY IS UNPREPARED WITH A DEFINITION EVEN THOUGH THE DISCUSSION ABOUT “CONCEPT CELL” LASTED ABOUT TWO MONTHS ON THIS SHORT LIST(per his claims). HE COULDN’T EXPLAIN THE “GRANDMOTHER “ISSUE HOWEVER HE KNEW THAT IS NOT “GRANDMOTHER” !
HAVE TO ANSWER IF THEY “HAVE A CLEAR, SPECIFIC DEFINITION OF THE CONCEPT CELL IDEA OR GRANDMOTHER CELLS”. ASIM ROY IS UNPREPARED WITH A DEFINITION EVEN THOUGH THE DISCUSSION ABOUT “CONCEPT CELL” LASTED ABOUT TWO MONTHS ON THIS SHORT LIST(per his claims). HE COULDN’T EXPLAIN THE “GRANDMOTHER “ISSUE HOWEVER HE KNEW THAT IS NOT “GRANDMOTHER” !Thu, Aug 18, 2011 at 12:15 PM Asim Roy:I have characterized it in a recent paper. But I don’t know of an exact definition.
AFTER MORE THAN 15 MIN Asim Roy CAME WITH KOCH’S GROUP DEFINITION
Thu, Aug 18, 2011 at 12:33 PM Asim Roy: I would simply say that concept cells are part of a very sparse representation and are highly informative.
QUIAN QUIROGA PROVIDES THE HISTORY OF THE ENTIRE ISSUE, NOT WITH A DEFINITION REGARDING GRANDMOTHER CELL:
Thu, Aug 18, 2011, at 1:07 PM Quian Quiroga: Ask 5 people about the definition of grandmother cell, you get 6 answers...
Many misunderstandings about this issue actually comes from using different definitions. The term grandmother cells comes from Jerry Letvin in the 60's but the idea was also discussed by Konorski (gnostic cells) and goes back to Sherrington and even James. Letvin, in his famous parable at an MIT course claimed that a (fictitious) surgeon ablated ~18,000 neurons coding for 'mother' and then eradicated the concept. Since 'mother' was not very objective then he went for the 'grandmother neurons' and hence the name ... As far as I know, Letvin didn't say that these neurons encoded for one and only one concept (grandma), but this was somehow implicit and it is what people typically assume.
Here are some definitions:
1) One and only one neuron encodes one and only one concept. This is of course impossible, at least for our data. If we do find a neuron firing to Jennifer Aniston, there have to be more.
2) Many neurons (and 18,000 seems like a good estimate) encode for one and only one concept. This is in principle possible, but... the Jennifer Aniston neuron also fired to Lisa Kudrow in a following session (it's in the suppl. material of the Nature 2005 paper); another neuron fired to Luke Skywalker and Yoda, and there are many other examples of responses to more than one concept (many of them in a Curr Biol. 2008 paper). Moreover, if we do find a neuron firing to only one concept, maybe it could have also fired to something we didn't show in the experiment.
One can still argue that these cells still encode single concepts (the blonde girls from "Friends", the Jedi from Starwars), but it is a discussion about semantics (after all I could say that neuron A fires only to the concept: all the things that makes neuron A fire). The important fact is that these neurons fire to concepts (perhaps more than one and they are typically related) and not to details of a particular picture or stimulus.
3) Many neurons fire to a few concepts (but not details of a particular picture). Well, you can call this sparse (invariant and explicit) representation, and that's what we did (TiCS 2008).
4) Other people will call an abstract (conceptual) representation grandmother cell coding.
Pick up a definition and call them as you like, but for me the important thing is that these neurons have a sparse, invariant and explicit representation, which is ideal for the medial temporal lobe memory functions.
Many misunderstandings about this issue actually comes from using different definitions. The term grandmother cells comes from Jerry Letvin in the 60's but the idea was also discussed by Konorski (gnostic cells) and goes back to Sherrington and even James. Letvin, in his famous parable at an MIT course claimed that a (fictitious) surgeon ablated ~18,000 neurons coding for 'mother' and then eradicated the concept. Since 'mother' was not very objective then he went for the 'grandmother neurons' and hence the name ... As far as I know, Letvin didn't say that these neurons encoded for one and only one concept (grandma), but this was somehow implicit and it is what people typically assume.
Here are some definitions:
1) One and only one neuron encodes one and only one concept. This is of course impossible, at least for our data. If we do find a neuron firing to Jennifer Aniston, there have to be more.
2) Many neurons (and 18,000 seems like a good estimate) encode for one and only one concept. This is in principle possible, but... the Jennifer Aniston neuron also fired to Lisa Kudrow in a following session (it's in the suppl. material of the Nature 2005 paper); another neuron fired to Luke Skywalker and Yoda, and there are many other examples of responses to more than one concept (many of them in a Curr Biol. 2008 paper). Moreover, if we do find a neuron firing to only one concept, maybe it could have also fired to something we didn't show in the experiment.
One can still argue that these cells still encode single concepts (the blonde girls from "Friends", the Jedi from Starwars), but it is a discussion about semantics (after all I could say that neuron A fires only to the concept: all the things that makes neuron A fire). The important fact is that these neurons fire to concepts (perhaps more than one and they are typically related) and not to details of a particular picture or stimulus.
3) Many neurons fire to a few concepts (but not details of a particular picture). Well, you can call this sparse (invariant and explicit) representation, and that's what we did (TiCS 2008).
4) Other people will call an abstract (conceptual) representation grandmother cell coding.
Pick up a definition and call them as you like, but for me the important thing is that these neurons have a sparse, invariant and explicit representation, which is ideal for the medial temporal lobe memory functions.
18, 2011 2:13 PM Dorian Aur:The firing rate in single cells does not have any reliable information regarding semantics --see the counterexample http://neuroelectrodynamics.blogspot.com/p/spike-directivity.html . The frequency rate not having any information was written by Asim several times to highlight the difference between views.
I've asked Asim to see the subtle difference between our statements
If thousand of neurons are needed to provide the concept then the "concept neuron" definition is confusing , the "expert" neuron term (Graybiel MIT lab) fits better the behavior of these neurons. (e.g not sure if the name (concept, place, expert, mirror) tells anything. The meaning is just for us --remark from Dr. Freeman).
I've asked Asim to see the subtle difference between our statements
If thousand of neurons are needed to provide the concept then the "concept neuron" definition is confusing , the "expert" neuron term (Graybiel MIT lab) fits better the behavior of these neurons. (e.g not sure if the name (concept, place, expert, mirror) tells anything. The meaning is just for us --remark from Dr. Freeman).
Thu, Aug 18, 2011 at 8:16 PM Asim Roy , when did I say that: “The frequency rate not having any information was written by Asim several times to highlight the difference between views.” You must have misunderstood. Thu, Aug 18, 2011 at 8:28 PM Asim Roy: I am reposting what I said earlier regarding your statement: “Therefore, Dr Koch knows well the trick : the patient can control the firing rate and modulate the image on the display using the feedback (it was revealed by many others before and is common all three above examples) . The paper does not demonstrate the concept cell idea!!!”
They found those four concept cells, corresponding to the four different objects, prior to the actual experiment. Moran discussed that in his presentation at IJCNN and you were there. And it’s in their paper. So it does demonstrate the concept cell idea.
ON FRI, AUG 19, 2011 AT 10:45 AM, I HAVE CHANGED THE TOPIC TITLE WITH A DIFFERENT ONE SUBJECT: THE UNRELIABILITY OF FIRING RATE
On Fri, Aug 19, 2011 at 10:45 AM Dorian Aur: To summarize the previous discussion:
(i)Quiroga et al. 2007 have "shown the concept" relationship not Moran Cerf (see Dr. Koch statement!)
(ii)The "concept neuron" proposal by Moran Cerf group is similar to the gnostic neuron model presented by Konorski in the early 1960s (La même Jeannette autrement coiffée)
1. First there is no "Jennifer Aniston" (JA) neuron. The same neuron has to respond to many other presented objects (otherwise is a grandmother cell) . Therefore, Quiroga et al., have extensively filtered the data to obtain this outcome (Quiroga et al. 2005) --details can be provided. When you record 10^3 neurons, have complex random temporal patterns and use statistics with an ambiguous measure (firing rate) everything becomes possible
2. Even when this neuron responds to JA it may in fact respond to different other features that are present in those images, however the firing rate measure is not sensitive enough to detect subtle aspects
3. Very important a strong firing rate to JA may indicate a reorganization triggered when JA is repeatedly presented not JA concept
4. To get reliable semantics from experimental data a direct relationship with "memory" has to be extracted . Spike directivity provides directly this relationship since it relates specific information with the topography of analyzed neuron . See the difference between the spider presentation and JA presentation (http://neuroelectrodynamics.blogspot.com/p/spike-directivity.html) specific parts of the neuron are active . Based on firing rate one cannot distinguish between spider and JA (8Hz ) in both cases
5. The reorganization always occurs in repetitive tasks and can change the neuronal response. It is unlikely to extract from an experiment the entire "blond women" (the concept) from a single neuron.
(i)Quiroga et al. 2007 have "shown the concept" relationship not Moran Cerf (see Dr. Koch statement!)
(ii)The "concept neuron" proposal by Moran Cerf group is similar to the gnostic neuron model presented by Konorski in the early 1960s (La même Jeannette autrement coiffée)
1. First there is no "Jennifer Aniston" (JA) neuron. The same neuron has to respond to many other presented objects (otherwise is a grandmother cell) . Therefore, Quiroga et al., have extensively filtered the data to obtain this outcome (Quiroga et al. 2005) --details can be provided. When you record 10^3 neurons, have complex random temporal patterns and use statistics with an ambiguous measure (firing rate) everything becomes possible
2. Even when this neuron responds to JA it may in fact respond to different other features that are present in those images, however the firing rate measure is not sensitive enough to detect subtle aspects
3. Very important a strong firing rate to JA may indicate a reorganization triggered when JA is repeatedly presented not JA concept
4. To get reliable semantics from experimental data a direct relationship with "memory" has to be extracted . Spike directivity provides directly this relationship since it relates specific information with the topography of analyzed neuron . See the difference between the spider presentation and JA presentation (http://neuroelectrodynamics.blogspot.com/p/spike-directivity.html) specific parts of the neuron are active . Based on firing rate one cannot distinguish between spider and JA (8Hz ) in both cases
5. The reorganization always occurs in repetitive tasks and can change the neuronal response. It is unlikely to extract from an experiment the entire "blond women" (the concept) from a single neuron.
AFTER 15 MIN ASIM ROY POSTED THIS MESSAGE ON THIS NEW TOPIC TO STOP THE NEW DEBATE CLAIMING ISSUES REGARDING ETHICS AND ALSO THAT MORAN’S TALK WAS INTERRUPTED AT IJCNN (LONG TIME BEFORE THIS DEBATE)
Fri, Aug 19, 2011 at 11:01 AM: Asim Roy: Please don’t respond to Dorian Aur anymore. You never know when he is going to post your comments on some public mailing list without your permission. He has no ethics. And I think we have had enough of a discussion on his theory that firing rates and spike timing interval has no meaning. I have repeatedly asked him to explain the results of Moran Cerf experiments using his theory and he is unable to do so. I think we had enough.By the way, for those who were not at the IJCNN talk by Moran Cerf, Dorian Aur interrupted Moran’s talk, took out his IJCNN poster and started talking. I had to stop him after a while so that Moran could continue.ANOTHER NEW TOPIC WAS BORN “CONCEPT NEURONS - WHAT IS A CONCEPT” Author ASHER EVANS
Fri, Aug 19, 2011 at 2:16 PM ASHER EVANS Prof. Quiroga said, "One can still argue that these cells still encode single concepts (the blonde girls from "Friends", the Jedi from Starwars), but it is a discussion about semantics (after all I could say that neuron A fires only to the concept: all the things that makes neuron A fire)."
To me, this raises a critical point and question. What does it mean to say that neuron, A, fires in response to a concept? Let f(A) be the set of stimuli to which A significantly responds, which may best be regarded as a fuzzy set (see Zadeh 1965 for treatment). If f(A)'s membership is very arbitrary, including the moon, two dissimilar species of beetle, and broccoli, among a host of other completely unrelated items, informative as A's response might be, it would seem A ought not be called a concept cell. But if A responds only to blonde females in the cast of Friends, it seems it would be more appropriate to call it a concept cell. In one case, "all the things that make neuron A fire" seems to correspond with a concept. In the other case, it doesn't seem to. What's the difference?
.......
To me, this raises a critical point and question. What does it mean to say that neuron, A, fires in response to a concept? Let f(A) be the set of stimuli to which A significantly responds, which may best be regarded as a fuzzy set (see Zadeh 1965 for treatment). If f(A)'s membership is very arbitrary, including the moon, two dissimilar species of beetle, and broccoli, among a host of other completely unrelated items, informative as A's response might be, it would seem A ought not be called a concept cell. But if A responds only to blonde females in the cast of Friends, it seems it would be more appropriate to call it a concept cell. In one case, "all the things that make neuron A fire" seems to correspond with a concept. In the other case, it doesn't seem to. What's the difference?
.......
on Sat, Aug 20, 2011 at 10:53 AM Dorian Aur: Yes, yes... excellent you've discovered yourself the truth about the "concept" cell framework. You do not need to take for granted any other claims, doesn't matter where they come from. In Science you need to find the truth by yourself.(1) If the cell fires only for "Jennifer Aniston" then it is a grandmother cell --- this model is not reinforced by any reliable analysis of experimental data
(2) If the cell responds to "the moon, two dissimilar species of beetle and broccoli" we cannot call it concept cell , however that's the correct experimental outcome in a correct analysis of data http://neuroelectrodynamics.blogspot.com/p/spike-directivity.html
(3) Our general understanding of concept points to a cell that responds only to "blonde females" however this basic idea reinforces the problem of grandmother neuron.
Unless Quiroga et al provide a novel definition for the "concept" framework then from
(1), (2) and (3) results that the problem of concept cell is ill posed.
Q.E.D.
The entire propaganda is meant to keep alive the spike timing dogma. You should read Neuroelectrodynamics (NED) and you'll understand more about brain computations. A short introduction is in the paper "From Neuroelectrodynamics to Thinking Machines" www.springerlink.com/index/X1L7388475323758.pdf
I'll be very pleased to answer any questions regarding NED, I'll always encourage a fair scientific debate.
"It doesn't matter how beautiful your theory is, it doesn't matter how smart you are (or how many years you can keep a false theory alive !) If it doesn't agree with experiment, it's wrong."
Richard Feynman
