Dear Joaquín,

In my opinion, synaptic and structural plasticity cannot be excluded. Emergent activity from the circuits is influencing reciprocaly the underlying structure and forth -see for example neurorehabilitation. There is evidence where structure can explain function itself, but this structure is plastic, so it is in constant modification. So, we can only speak about a connectome of an instant of time, rather than in a continuous period of time.

Other recently observed phenomenons are adding difficulty to understand the working of cognition besides structural circuitry and the emergent function through classic understanding via membrane conductances, such as the effect of electric fields and volume conduction of activity coming from distant non-axonal propagation.

I guess the puzzle is of increasing difficulty as soon as more understanding is acquired from the previous stage, even though I don't have all the understanding -a fact that torturates me constantly :)

While the fine connections morph, at the gross level there is consistency in the links between areas suggesting that at the network level the connectome is consistent. This means that the connectome is based on uncertainty, but that there is an architecture to the networks that make up the brain.

The answer is "no". A map of neuronal connections alone does not reveal the essential structure and dynamics of the particular kinds of neuronal mechanisms that do the cognitive work.

I presume that the reference to the genome is related to the disappointment that decoding the genome did not explain all of the problems of mankind.  I think that the connectome has most clearly demonstrated that old-fashion neuroanatomical connections are valid.  Understanding the connections more clearly, more connections that are uniquely human, and the variations in the weight of the connections between individuals and relationships to disease conditions and disorders (Alzheimer's disease, traumatic brain injury) will be of great use.  However, the actual details of thoughts are still buried in the complexity of the microscopic connections within the cortical neuropil, under continuous modification of numerous subcortical systems.  My conception of this (Ashford, Coburn, Fuster, 1998) is that the neurons of each cortical region work as a dynamic computer, with the pattern of neuron activations at any one instant providing an analysis with the power of a multi-million (number of local pyramidal neurons) -bit computer (with the biggest man-made computers usually being about 64-bit).  The number of unique states would be 2 to the multi-million power.  The thoughts are buried in this nearly infinite (much more than all the particles of the universe) complexity, with the quadrillion synaptic connections of the brain constantly changing, at a rate of hundreds of millions per second when information is being encoded.

There is an old story about aging and dendritic length, which I am trying to revisit at the moment.  Buell and Coleman (1979) argued that dendrites get progressively longer with age.  But this was disputed, at least for Wernicke's area, by Jacobs and Scheibel (1993), arguing that aging dendrites get shorter.  But, it makes sense that, with their journey through a long life, dendrites would continue to get continually longer, as much as is humanly possible, constantly search for better connections, seeking the true path.  So, an individual’s real perspective of the world lies in the essentially infinite possible connections of the dendrites of the 100 billion neurons of the brain, the ultimate, unknowable connectome.

J. Wesson,

With such a vast search space we have to formulate and test theoretical models of competent cognitive mechanisms. For example see the SMTT experiment in "Space, self, and the theater of consciousness" on my RG page.

Dear J. Wesson! You very elegantly and lucidly explain complex things. Possibilities of neural networks of the human brain are infinite and poorly studied. An infinite number of connections between brain cells is the basis of diversity of feelings, thoughts and creations. You concisely drew attention to the possibility of even a single neuron, which is able to process information, to remember, to compare, to make decisions and monitor the implementation of the plan. For example, aware of the unique capabilities of even a single neuron of pedal ganglion snail. This is only a dream artificial Connectome. What to say about the Universe of interconnected neurons in the Human brain!

It all depends on what you mean by "cortical dynamics of cognition". If this is the exact patterns of activity of many neurons plus the knowledge of what they process and how, knowing the connectivity would not hurt, although advances in recording methods would be required to achieve the goal. 

"Cortical dynamics of cognition" - this productive "conversation" cells of the cerebral cortex with each other, which is accompanied by the emergence of awareness of themselves and others phenomena (cognition). We are in reality only partially aware of the events in the cerebral cortex. Our fragmentary knowledge of human cognitive processes prevents us formulate the design of artificial intelligence.

On the other hand, ignorance is an incentive for learning and creativity.

The problem of unawareness of some brain processes can be solved using neurofeedback. (Not that I fully understood your definition of "cortical dynamics of cognition"...)

Dear Mikhail! And who really understand all the subtleties of the functioning of the neural networks of the human brain? (excluding the Creator)

May be Globus Pallidus because it has many converging inputs :)

Dear all,

            I have been following with interest your answers to my question.  To further clarify that question, it seems I should add that by “the cortical dynamics of cognition” I mean the neural mechanisms enabling the cerebral cortex to use memory and knowledge in cognitive functions: attention, perception, memory formation and retrieval, working memory, intelligence and language.  Although my focus is on the cortex as the prime substrate of cognitive networks, I do not exclude from those mechanisms the limbic system, the brainstem, the diencephalon or the basal ganglia.

             To the best of my knowledge, the answer to my question is no: the connectome by itself will never explain or predict the dynamics of cortical cognition.  This opinion, which is certainly not intended as the last word in this discussion, is based on the following considerations:

1.  The connectome defines structure, not dynamics.  It is a stupendous refinement of tractography, which has been around for a long time, though never before with the precision allowed by modern imaging.  However, here function cannot be inferred from structure, for one thing because cognitive function has temporal constraints that the connectome, a static structure, does not have.

2.  The fluidity and adaptability of cognition do not depend on nerve-fiber connectivity alone, no matter how exhaustively exposed, but on myriad synaptic transactions that impart dynamics to that connectivity and are inaccessible to imaging methods.

3.  Those transactions take place in synapses whose weights are constantly changing, even in sleep, and to which the connectome is blind.   Here the fact that some cognitive functions take place unconsciously is beside the point—though phenomenology and psychophysics are invaluable methods in cognitive neuroscience.

4.  It seems to me unimaginable that the connectome could ever “dissect” a piece of active knowledge or memory within an immense and ever changing system of overlapping, nested and interconnected networks, where a neuron or group of neurons can be part of many memories or items of knowledge. (After proprietary delays, my conceptual and empirical review of the subject is finally accessible in RG, “Cortex and memory: emergence of a new paradigm.”  J. Cognit. Neurosci., 21:2047-2072, 2009.)

            So far, most all of you seem to share that opinion.  Does all that mean that the connectome is a fruitless enterprise?  Not at all.  Very appropriately, Wes advances reasonable ideas of some of its potential applications to medicine.   Further, some of us evidently envision how, in combination with other methods (e.g., electrophysiology, DTI), the connectome can be helpful in clarifying some aspects of the neural dynamics of cognitive function.

 Cheers, Joaquín     

From my research, cognition is a combination of connectomics and neuron physiology. One of the prevalent aspects of cognition is not only the temporal aspects of the individual neurons but the temporal aspects of the connectome overall. We often see the same neuron being activated with different resulting behaviors based on what set of neurons were involved in the activation and when. Researchers often look at the neocortex as the decider of behavior but the neocortex is only trying to influence behavior by manipulating the activation of specific cortical regions through the connectome. The connectome has a lot of prewired functions and the way we change that function, if we can change it, is by "flooding" that region with connectomic influence.

Also we have discovered, adding to the above, is that behaviors are highly integrated within the connectome so one cannot "remove" a behavior without affecting many other behaviors. For example, if we ablate one significant neuron in the C elegans connectome, a neuron as noted above that can be activated for many different types of behaviors, we can change the way the worm behaves entirely and one that may not allow it to survive for very long.

Sorry, there is much more but I am pressed for time at this moment,

Tim Busbice

Dear Tim,

         Your answer is truly amazing, because it gives us vivid evidence that cognition in the nervous system of a nematode operates according to the very same principles as cortical cognition in the mammalian brain.  To wit, using your words:

1.   “...cognition is a combination of connectomics and neuron physiology.”  Indeed, cognition does not result from connections alone but from the activity of the connected neurons (cf. my points #1 and #2).

2.   “…not only the temporal aspects of the individual neurons but the temporal aspects of the connectome overall…“ (cf. my points #1 and #4).

3.   “…the same neuron being activated with different resulting behaviors based on what set of neurons were activated”  (cf., my point #4).

4.   “The connectome has a lot of prewired functions…”  So do the primary sensory and motor systems in the mammalian brain.  They are part of what I call “phyletic memory,” the memory of the species, with which we are born and which in its connective structure incorporates the modes of adaptation to the environment that the species has acquired in the “night of times” (evolution).  It is a form of memory, because it is stored—prewired--information, ready to be “recalled” or retrieved with every sensation and every motor act.

  5.  “…behaviors are highly integrated within the connectome so one cannot ‘remove’ a behavior without affecting many other behaviors.”  I could not say it any better for the executive cognitive networks of the mammalian cortex.

          So, your little critter is elegans in more ways than one!

 Cheers, Joaquín

Dear Joaquin:

Thank you for your kind response. I'm not sure about your point #1 though? I may have misinterpreted your meaning?

The connectome defines structure, not dynamics.

One of my points is that the Connectome does certainly define dynamics. Other experiments have shown changes in behavior of animals if we connect neurons that are not typically connected (http://www.ncbi.nlm.nih.gov/pmc/articles/PMC4588382/). The "routes" and more importantly, the timing aspects of those routes are what gives us cognition. I think a macro example of this could be when our emotions come into play, the timing/routing of our thought processing changes because of a set of neurons that heavily influences (stimulates) other regions of the brain and we do things that we probably would not do otherwise. 

I've been researching artificial connectomes for some time now and I can produce a number of cognitive processes using very simple neuronal models and by changing the connectivity aspects of a connectome, deliver many different results. There is a very clear and thin line of connectomics in that too little or too much "connectivity" and how those paths are configured, can render a connectome useless.

I am developing artificial connectomes to create cognitive (connectomic) AI and the reason I have moved to artificial connectomes this past year is because animal connectomes are too prewired to make use in the development of intelligent computer systems, although there are very good reasons to emulate animal connectomes and I have been able to do so. The keys to using animal connectomes in the development of intelligent systems is deciding what sensory neurons to stimulate for a given situation and how to interpret the motor neuron/muscle output? Even though one may get results not desired or expected when using animal connectomes for intelligent computer systems due to the prewired nature, I find that using animal connectomes gives better results than most current AI technologies such as Deep Learning or ANN type systems. 

Thank you for bringing up a great topic and obviously one that is close to my heart,

Tim Busbice

Dear Tim:

            You are right.  My statement about a connective structure devoid of dynamics was ambiguous if not incorrect.   It is plainly contradicted by my later statement on the prewired nature of sensory and motor systems, which are genetically inherited, yet very much contain function.   There is no doubt that the connectivity of those systems defines and, under given circumstances, determines their function.  It seems that the most plausible AI models and systems are based on, and even simulate, animal sensory and motor systems.  This is not incompatible with their capacity to develop emergent properties and functions.   That capacity is most definitely present, over and above prewired structure, in associative sensory and motor cortices.  Without it, there would not be either imagination or planning in our lives.

            Otherwise, I stand by my statements and the admiration of your work.

 Kind regards, Joaquín

Dear All,

Very interesting discussions! Since I have recently started to formulate neuron dynamics from a chemical physics viewpoint, it is important to make the distinction between structure and dynamics. Nevertheless it makes a fundamental difference whether the structure is classical or not.

However a quantum version of the dichotomy structure-dynamics is presently controversial as recent discussions, i.e. Tegmark, Penrose Hameroff, Bernroider etc., focus on the possibility to overcome the QM decoherence problem.

Recent advances in chemical physics do, however, formulate open systems dynamics consistently in that it permits the evolution of proper time scales for the generation of information transfer cf. the genetic code, which might be of interest also in connection with cognition, see more details on my RG page.

Hence one might ask the question whether the concept of a connectome might be limited from the beginning or whether it might permit meaningful generalizations perhaps beyond the classical domain. 

Dear Erkki,

            Your answer to my question is extremely interesting.  The connectome, I think, indeed permits generalization beyond the classic domain.  This would be of much theoretical importance. 

            I must confess I have trouble reasoning about cognition at a sub-network level, but I keep an open mind, if nothing else because this allows us to confront questions of general neurobiology such as those you raise.  The root of my trouble is that I view all cognition in relational terms.  All cognitive information is encoded, stored, and retrieved in the bran in those terms.  Any item of knowledge, memory or perception is encoded by a set of relations within itself (a Gestalt) or between itself and other items.  After more than half a century of dealing with single neurons, intra- and extra-cellularly, I cannot conceive of any item of cognition encoded at e less than network (i.e. connective) level.  Mach’s “elementary unit” of sensation does not exist.  All sensation, however simple, is encoded by a set of irreducible relations.

            Having said that, I feel compelled to caution.  Many of the more or less controversial dualities that you mention evoke spurious dichotomies between phenomena at different parts of one and the same spectrum or different applications of a same general principle.  All of us have heard about “the special case” of this or that.  The most notable, of course, in your field of interest is that between Newtonian and quantum physics.  Now, a comment about the dynamic qualifiers of the connectome.

            Don’t we come to this world with little connectomes in our sensory and motor cortical areas?  In our very distant ancestors, light of the same spectral frequency, for example, must have arrived in certain parts of their visual system more often and at the same time than light of other frequencies, if nothing else because of the continuity and contiguity of colors in the objects of the physical world.  The same may have been true for other visual features.  Is it not conceivable that those patterns of relational contiguity and continuity led to the phylogenetic formation of feature detector networks in sensory--e.g., visual--“fields” of primary cortex?  They would have formed the genetically inherited structure of networks (connectomes if you will) ready to dynamically recall from phyletic memory the features of our environment with every perception and every motor act.  It is not difficult to imagine that the same principles of relational contiguity and continuity would apply to the plastic regions of an individual’s cortex.

Kind regards,  Joaquín

 Hayek, F.A. (1952). The Sensory Order. (Chicago: University of Chicago Press).

Thanks for your kind words Joaquin!

I value your fundamental insights of the world of neurons in particular since I am a relative newcomer into this fascinating field, hoping that my background in chemical physics might, with the aid of the undeserved fortune of an ignoramus, perhaps may cause a stumble on something meaningful.

For instance in order to promote cognition as a fundamental function of consciousness I vision the whole undertaking from (i) perception, (ii) awareness, (iii) cognition, and (iv) feelings and also emotions as an irreducible process of the mind, yet being possibly characterized by fundamental time scales.

Behind this list of steps subsist a “protracted communication hypothesis” that I find to be intrinsic to every life form from the microscopic genetic code to macroscopic domain. It originates already at the initial stage, i.e. about 3-7 ms, of the sensory neocortex-reorganised background due to the impact of e.g. a conditioned stimuli etc. The next stage concerns awareness, cf. Libet’s conscious awareness delay up to several hundred ms, which nevertheless retrospectively creates a subjective back track.

Hence the conscious mental experience cannot survive without the brain processes that give rise to it. It seems that the steps (i) perception, (ii) awareness can be separated according to specific time scales, i.e. (i) 5-10 ms, (ii) 200-500 ms, while (iii) cognition is fundamentally irreducible as regards the time scales of (i) and (ii) and possibly also within longer durations related to (iv).

The hypothesis above appears to be in good agreement with Arnold Trehub’s Retinoid System, although there are many pieces still to adjust and understand. I do not know if this makes sense in your cognition theory.

Best regards 

erkki

By the way, since there are such great scientists here, may I interest in contributing to this research topic in Frontiers:

http://journal.frontiersin.org/researchtopic/3099/paradigm-shifts-in-neuroscience