Simon: Visual experience is inseparable from sensory experience as a whole, which reflects external and internal phenomena. On an intuitive level it feels "real," but this impression is illusory. The phrase "taking in visual information" seems to reflect this intuitive naïveté. Electromagnetic radiation enters the eye; that which affects one's sense of sight may be called "light." "Visual information" is in the mind. You are right to question common metaphors like the ones you mention, as they tend to cloud our understanding of phenomenal experience.

This is commented on, discussed and debated in a book, actually a reprint of an issue of an academic journal which has become something of a classic, at least in some circles. You are probably aware of it, but I think the reference bears repeating because of its relevance to your question.

Is the visual world a grand illusion? (reprint of nos. 5 and 6, vol. 9, Journal of Consciousness Studies), Thorverton/Charlottesville, Imprint Academic/Imprint Academic Philosophy Documentation Center, 2002.

Table of contents (adapted from http://www.ingentaconnect.com/content/imp/jcs/2002/00000009/f0020005)

Is the visual world a grand illusion?

pp. 1-12

Author: Noe, A.

How could I be wrong? How wrong could I be?

pp. 13-16

Author: Dennett, D.

There is no stream of consciousness. What is all this? What is all this stuff around me; this stream of experiences that I seem to be having all the time?

pp. 17-28

Author: Blackmore, S.

The grand illusion and petit illusions: interactions of perception and sensory coding

pp. 29-34

Author: Bridgeman, B.

How well do we know our own conscious experience? The case of visual imagery

pp. 35-53

Author: Schwitzgebel, E.

Our perception of the world has to be an illusion

pp. 54-71

Author: Ballard, D.

Visual worlds: construction or reconstruction?

pp. 72-87

Authors: Davies, T.; Hoffman, D.; Rodriguez, A.M.

The Tinkerbell effect: motion perception and illusion

pp. 88-101

Author: Durgin, F.

Is the visual world a grand illusion? A response

pp. 102-110

Author: Mack, A.

Change blindness blindness as visual metacognition

pp. 111-130

Author: Levin, D.

Is visual experience rich or poor?

pp. 131-140

Author: Siewert, C.

The grand grand illusion illusion

pp. 141-157

Author: Cohen, J.

Two dogmas of consciousness

pp. 158-180

Author: Rowlands, M.

Is seeing all it seems? Action, reason and the grand illusion

pp. 181-202

Author: Clark, A.

David

thanks for the ref. I did not know it ( I know the work of many of the authors). It is now ordered :)

The orientation of many of these papers (titles anyway) alerts me to the fact that another way to frame my concern is in terms of the internalism/externalism debate. And this, in turn is an aspect of our general acceptance of Cartesian (mind/body) dualism, and this I am quite comfortable in rejecting entirely. Not that there is no 'mind' but that to speak of 'mind' as being separable from or complementary to 'body' is a nonsense. Sadly, it is a nonsense that most people subscribe to.

Yes, Simon. Once one breaks free of these aspects of our medieval European cultural heritage, it all seems pretty evident. Looking back, it has been a long, arduous personal journey, and it isn't surprising that this paradigm continues to flourish, in spite of advances on many fronts. I'm glad you found the reference useful.

Hi Simon,

A proper question and the line below works for me.

"Could it be that we find the brain mysterious in part because we apply inappropriate structuring metaphors which confound our inquiry?"

But actually we apply the wrong ontology, vision is prior-to science. Until we study the phenomenon as the phenomenon we won't make much progress. The notion that the visual pathways are segmenting a picture to then build it back up to perceive it prevents us even presenting the visual system with appropriate stimuli. 

The rot goes deeper?

Dear Simon,

It springs to mind that the whole issue packed into your question revolves around the word "information" and, necessarily, your personally meaningful conceptualization of what information is - at any rate, in the particular context of your framing your question. Different interpretations are possible and of course evident, reflecting different theoretical assumptions derived from the perspectives of different domains of inquiry - "information" in the context, e.g., of Shannonian theory, of signals transmitted through and between adjacent neurons, of conscious percepts upon which courses of action are decided, - without mentioning more mundane applications of the word.

Given my uncertainty as to how you yourself define "information", it's impossible for me to interpret your question in the way you wish it to be understood, and hence ask you to clarify. However, in your question's elaboration, your concerns about axioms and their confounding consequences for reasoning "veridically" about anything are stated most clearly. Your desire to abolish existing axioms leaves me, I admit, feeling it's unfair of me to request you to define "information", for, without you resorting to some axiomatic foundation upon which to formulate such a definition, surely, am I not asking the impossible?

In short, I believe (i.e,, assume axiomatically) that meaningful discourse, even of the most primitive kind, is impossible without a foundation of axioms, which I conceive as forms in which inborn aptitudes for constructing meaning actually manifest themselves and become operational. We cannot, even should we wish, abolish them, and were that possible, we would eliminate all that being human implies. Like Popper, however, I wholeheartedly acknowledge we can, and do well to, reject certain axioms and seek better ones that may be closer to veridicality, though we can never prove them to be. And, of course, this is what science, as a collective enterprise, over the course of decades and centuries, has continually sought to do and accomplished. But at any given instant, we have to fare the best we can on the ones we have.

Sincere regards and season's greetings!

Richard.

Richard,

thankyou for your thoughtful post. I do not aspire to abandoning all axiomatic assumptions. I do not see that as possible, to the extent that (all ?) knowledge is analogic. My quest is more in line with Popper's, as you describe it. My concern about 'information' regards the term in its computer-science sense. This appears to be the opertational definition, it tacit, on our culture generally and certainly in the academy.

What worries me (deeply) is that much science education seems to be indoctrination into a set of assumptions which together appear to constitute a coherent explanatory system. In contemporary neuro/cognitive/psychological worldview, 'information', 'information processing' and all sorts of notions from digital computing are transferred onto mind/brain, seemingly without interrogation.

I am quite familiar with Shannon (and von Neuman, and Turing, etc) but I'm not sure I can conceive of the act of looking/seeing as a passage of (visual) information. I'm waiting for someone to offer me a persuasive explanation for why I should think of seeing as a flow of information.

Are we left to choose between the radical externalism of Gibson - and solipsistic internalism? I cannot endorse the latter but I seem to be talking myself out of the former.

Hello

I have not any knowledge of visual or the brain.

From eye witness testimoney its seems that our sight is much comleceted than just taking in visual things. I think we built some patterns of things that we think that are in what we see

By Sdannon's original definition information is not simply input but rather it is a capacity of the reciver to receive (process) input. Visaul input is not sufficient to to explain eye movements: for instance saccades in the dark.Whch bits are selected as visual targets is thus partially determined by a pre=motor or para-motor plan see the literature on Pre-Emptive Perception ((bodis_wollner)   and also Alva Noe

Regarding computationalist metaphors - Rodney Brooks has a relevant short piece in Nature - posted at:

https://www.researchgate.net/publication/223138336_Turing_centenary_Is_the_brain_a_good_model_for_machine_intelligence

Article Turing centenary: Is the brain a good model for machine intelligence?

Ivan

 As Shannon said "The fundamental problem of communication is that of reproducing at one point, either exactly or approximately, a message selected at another point."  He was clear that this definition has nothing to do with semantic content. Note that his classic 'A Theory of Communication' established the 'bit' as the quantitative unit of 'information'.

Thankyou for the Pre-Emptive Perception references, I know Noe and ORegan's work but other refs are always welcome.

Dear Simon,

I have no difficulty empathizing with your position and stance!

However, after long decades of sporadically pondering the meaning of "information" - i.e., its context and role in the larger, evolving scheme of my personal theorizing (for which "coherence" of its ideas is the all-compelling objective) - I find myself nowadays able to conceptualize the notions of "information" and "information processing" in terms sufficiently reduced to allow me to generalize them, in bottom-up fashion, to the entirely alien contexts of telegraphic/radio transmission, the brain and the digital computer, without the need for invoking analogies that inevitably add unnecessary complexity to the interpreting of those terms and which invite irrelevant, fruitless and interminably distracting comparison-making between those inherently alien contexts. You seek an explanation that de-confounds the mystery and seeming incongruity of information and the brain's manifested abilities - which I maintain results from our instinctive and educationally nurtured reliance upon analogy and comparison as "default" intellectual tactics. Perhaps sharing my definitions here might provide foundations for an explanation to some extent?

Information, for me, means purely and simply, signals - streams of energy - that are relayed to something and cause a change in its state. The term involves no analogizing: information is not "like" a stream of energy; it is, by its very nature, streamed energy that has the opportunity to act on something responsive to it. The verbs "to process" and "to transform" are synonymous in my view, and hence the changed, or transformed, state of something, effectuated by a stream of energy impacting on it, constitutes processing of that energy, by virtue of its tranfer to the body receiving it. The recipient body's transformed state is manifested as the conversion of its matter to accumulating potential, or immediately released, energy, such that it "performs" an operation - the creating of a stream of energy having the capacity to "inform", and so "transform" the state of, another body in turn. The operation performed by the body can be conceptualized as the completion of its transformed state, thus defining the notion of "processing". (I do not hesitate in attributing my depiction, in part, to my incurable delectation for pondering words combining  a common root with different prepositions in search of greater conceptual understanding!  But I am again and again bowled over by the perspicacity of their inventors - so often individuals of Classical times - but ordinarily concealed by our un-analytical, habitual use of such wprds in everyday situations.)

The reiterative formula inform > transform > perform (> inform . . .) can be generalized, without alteration, to any device providing a medium for relaying streams of energy to other responsive components in controlled ways. Such media can be as alien in constitution as radio waves relayed between a transmitter and receiver, telegraphy networks, brains or digital computers. The formula is implemented by its elementary components - in neurons, for example, by the sequence of accumulating charge in dendrites, discharge and propogation of action-potentials, and resulting expulsion of neurotransmitters at presynaptic boutons. Or equally in the conceptual constituents of any simple machine - an interface for receiving energetic input, an actuator that transforms that energy, and a means of emitting the transformed energy as target-directed output. The physical capacity to process information, as so described, is the sole common requirement for the different media. No analogies or comparisons between them need to be entertained; each does what it does in its own idiosyncratic manner, prescribed and governed by the particular physical properties of its medium.

Per se, the information processed by any individual component of a multicomponent system cannot be considered in any sense interesting or meaningful. It is what it is and nothing more than. Interesting, and meaninginful transformations emerge, however, when information from diverse sources converges upon a second component. The emergent effect, or potential for the eventual expression of their now integrated properties, is passed on - again as seemingly uninteristing unitary signals - to other components that may also be receiving similarly composite, emergent signals, and what also emerges is the possibility of components controlling each other's activity directly or, with further convergent steps, indirectly, in the form of local and network-scale feedback loops. Eventually what may emerge might be the conscious experience of a visual scene, or alternatively of a language-encoded image such as a metaphor. The important points to note are, 1) that the emergent products are the products of particular, unique patterns of connectivity (ie., convergences of multiple sources that each have other divergent targets); and 2) that no emergent product, by virtue of its complex substrate, may be reduced or reversed to the informational properties of its substrate's individual components; the progress of information-processing is always uni-directional.

This depiction must, and I  believe does, suffice. As a basic working model or principle, it works for me - at the moment, at any rate. By implication of its avoidance of analogy, it permits me to conceive further, more specific models by way of a purely constructive thought-process uncomplicated and unconstrained by the influence of analogies I formerly resorted to. I say this particularly with your words "(? all) knowledge is analogic" in mind. From the standpoints, equally, of my personal experience and non-conformist theoretical understanding, I can only concur with your question-mark. All knowledge, in its original formulation (not only in early infancy but subsequently, too) is "concrete" in nature and, by nature, constrained solely by the context (environmental and emotion-encoded) in which it is perceived. Only as the result of its subsequent generalization (i.e. cognitive categorization and instantiation in other contexts) does analogy become possible, as a more internally and imaginatively conducted process. That is necessarily the order of progression even when perceptual processes are being selectively controlled by high-level analogical knowledge-representations. I have another dependable case in point. As a long-seasoned musician, when engaging practically, I think exclusively in terms of concrete sounds I wish to hear emerging from the piano; and so long as they do so, there is no need for me to analogize those sounds in terms of muscles, kinematics, trajectories, fingerings, visualizations of key locations, notational symbols or anything else of an extra-auditory nature. The properties of successive sounds, mentally imagined and then acoustically realized, are sufficient, since in reality instrumental music consists of nothing else. This is not in any way extraordinary; do you resort yourself to such analogizing your intention to speak in terms of the physical mechanisms involved in doing so, or likewise as you write, walk, drink from a cup and wash it up, or focus your eyes on a scene and "take it in"? The answer, I hope, won't be necessary. The brain does all this automatically as the result of having learned how to appropriately control the information-processing operations required for transforming intention into action.

By way of a parting (and admittedly slightly impish) thought I wonder, given the energy/matter conversion required by my account of information processing, whether your happiness to do away with Cartesian dualism might be tempered by the possibility that its formulation is simply misconceived and in need of correction. What I am referring to, of course, is Descartes' notion that bodies are physically extended and mental states have no such extension. My homespun account, if used to recast Cartesian dualism in terms of Einsteinian equivalence, implies roughly the converse; the extension of the physical medium allowing information processing is strictly dependent on its quantity of matter and the number of components that that quantity allows for, whereas the possibilities for dispersing (extending) energy converted from that matter is limited only by the number of possible connections (and, therefore, information-processing operations) allowed by the summed matter of its components. Assuming calculations of the number of neurons contained by the matter of the brain (c. 80 billion) and a (possibly conservative) estimated average number of connections made by a single neuron (10,000) the number of unique connection schemes possible works out at around 1080 . That, I think, is not too constricting as a scientific notion, since the number of information-processing events a person might clock up by age 100, assuming one roughly every 3 seconds, is utterly infinitesimal by comparison.

Anyway, be it sense or nonsense, that's my offering for now. Please, whenever it suits, tell me your reactions, or don't hesitate trying to pick my brain or challenge it as you might feel inclined.

Best regards!

I guess this is relevant here? 

The last painting of 2015. Lets hope things improve in 16!

4 fixations and 3 saccades. (overlaid)

Fixations are made in terms of subjective perceptual structure, i.e., within Vision-Space and not in terms of the geometry of optical projection or central perspective. So a record made of the experiential space that includes all the fixations does not 'add-up' in terms of environmental space. They do not amount to some navigation of a picture or even to the positions existing in the real setting. Spatial awareness is dependent on the navigation of the egocentric, condensed experiential space of the phenomenon.

The various branches/stalks of the plant belonging to the 3 individual fixations made on the flower heads would converge at the pot. However, they do not arrive in the same place in the overlaid experiential records. In fact the pot has to be referenced in 4 position, one for each fixation. The same is true for all the contextual settings that surround the various fixations. I have indicated these in drawn colour lines.

Why don't we notice all these 'changes' taking place as we make fixations? Because we don't make any records! There are no frames. The ONLY consistent spatial referencing system we experience is that of perceptual structure, the condensed field of Vision-Space. We don't 'map' the environment we experience the environment in terms of the subjective space we generate.

Further info www.pacentre.org

Best - J

Take a look at my TINS, 2015 paper. You'll find that I share your distaste for the brain being a wonderful computer metaphor.

Hi Moshe, This paper look very interesting. I read a paper by Joel Norman: Two visual systems and two theories of perception: An attempt to reconcile the constructivist and ecological approaches 2002. You guys in Haifa are involved in some interesting stuff! I will read your paper in detail. I will get back to you on it if I may? You may like to check out some material at www.pacentre.org?

All - just catching up.

Richard, thankyou for your extensive reply. I will read it again and think about it before I venture a response. Moshe, thankyou for the paper. John - can you send a link to the Norman paper? 

Regarding your closing sentence "We don't 'map' the environment we experience the environment in terms of the subjective space we generate." I completely agree. This is sometimes called Diectic perspective. It applies as much in linguistics, robotics and micronesian navigation. The western 'objective', cartographic or 'gods-eye view is a highly sophisticated cultural construction, just like perspective. In fact, I think they are isomorphic but I don't have the math to demonstrate it :) The part I find wanting in your accounts is the sensorimotor - I do not think we can speak of 'vision'. or at least 'seeing' without reference to learned and current non-visual sensory corelation.

happy new year to all

Norman's paper attached. 

The point about the dynamics of vision is that I think our perceptual structure and hence umwelt is heavily reliant on it. Our relationship to light drives the dynamical system of perception. I.e. other senses are perceived through the structure that our relationship to light impose on our perceptual structure. The other way to view that is to say that it's through the implicit visual field that we are able to accomplish multi-sense integration? I have precisely 'zip' proof of this. It just makes a shed load of sense to me. We generate a 360 degree field structure the locus of which we can control via attention. The default setting of the field is 'body centred' but we have also learned how to centre the field around a focal point in space, hence we (those with typical perceptual structures) are able to see ourselves in relation to the world. In both applications we (body, auditory, olfactory, and nervous system) are factored into the field structure. If we can can model this perceptual structure artificially, we should be able to deploy auditory input etc. into it as we do with visual input. Perceptually structured media? First we have to understand how this 'covert' processing system works! I think Vision-Space illustrate the output of this system but it will require a multidisciplinary team to understand just how it is that a neural cascade from our sense organs can populate the system. This may be 'difficult' but the 'difficulty' is not linked to the so called 'hard problem'. The system will demonstrate that there isn't one!

Well, walking through the door frame into 16 has done nothing for my brain's ability to remember stuff! Article is attached.

Thanks, John, for putting me onto the Norman paper!  My own work, aimed at theorizing the processing-architecture underlying exceptional expertise in music-performance,  recognizes as crucial the comparable 2-system, ventral-dorsal auditory pathway relationship - and the affective, executive and error-montoring mechanisms that allow co-ordinated, task-relevant linkage of concurrent information within the two pathways - for achieving effective percept-action (ie. ideo-motor) couplings, and therefore for explaining the (necessarily enactive, embodied) process of transforming intention into action (and vice-versa).

As an overview and sizing-up of the two systems' relevance to each other, the article is going to be a most valuable asset for me. Greatly appreciated, John!

Kind regards and wishes for the New Year!

Pleasure Richard.

I have taken an interest in perceptual processing on from the experiential approach. i.e. by studying the actual structure of the phenomenon (as it occurs to us) and then looking through the sciences to see what concurs and what doesn't. The arrogance is that I use what doesn't, to illustrate just what a mess the following of an approach lead by 3rd party observation (inappropriate ontology in my view) can get you in!! It's not all just theory however. I have made an attempt to deploy the basics of VIsion-Space in the form of a software package that can take optical record (pictures) in, reconfigure them to accord with perceptual structure and then batch these into perceptually structured media (moving images.) I.e. we have a basic architecture capable of modelling visual awareness. In my view there will be a very great deal of overlap with audition and vision with respect to perceptual structure. What goes on the superior colliculus and thalamic regions (old brain) especially. 

I have also made an archive of the relevant papers that I could give you an invite to if you like? Drop box. A key computational vision scientist is Prof Jan Koenderink. He has some very interesting material 'out there' and is one of the few scientists that understand the limitations of his discipline's ontology (that vision is prior-to science) and so has been able to develop ways to see a bit further. VIsion-Space deploys some of his insights.

Are you aware of the work of Bregman (Auditory scene analysis) and Plomp with respect to audition? Brilliant stuff. 

Do check out the articles on my RG profile and the VS presentations. The web site being www.pacentre.org. 

Thanks you Simon for posting these relevant questions. Very best of luck with your conference in 16. 

By the way the programming code for Vision-Space is available 'open source ' from Cardiff Uni (computer science and informatics). They are trying to obtain funding to get Vision-Space working at realtime speeds for a simulator. Using an eye tracker we should be able to get individuals making realistic (in perpetual terms) 'fixations'. We can then get reliable evaluation data out of the new form of illusionary space. Of course we will need an audio input to such a system! The auditory system also uses the same attractor (explicit) and contextual (implicit) forms of attention. 

Further thanks to you, John!

Fascinating and inspiring to learn of your work, and a first glance at your papers on RG promises some most edifying reading in store for me.

And, in reply to your hugely generous invitation to view your archive, I feel thoroughly honoured! My imagination tells me it contains a cornucopia of perspectives I haven't entertained, and, glutton that I am for anything that might add to my own limited set, how could I refuse? I'm deeply grateful.

Regarding Bregman - yes, I'm acquainted with some of his papers and their broad import; but Reinier Plomp's work is new to me. I've just accessed an interesting-looking paper of his on consonance which is going to be one of this evening's reads.

My reiterated thanks, for all of the above!

Simon,

Have you run across the following Frontiers research topic? Possibly its articles might interest you:

http://journal.frontiersin.org/researchtopic/1713/radical-embodied-cognitive-neuroscience#articles

Happy New Year!

Hi Simon. 

Richard has been through some of the the publications collected around Vision-Space and thought that you might be interested in the attached by Prof. Jan Koenderink. This was sent to me in a draft form prior to publication and I have no idea of the link or even if it was published. I think it was to be a chapter in a book? Anyway the title is a give away! 

vision is looking for gestalt.

For info:

The APA citation for the Koenderink paper John posted: 

Koenderink, J. J. (2010). Vision and Information. in: Perception beyond inference: The information content of visual processes, p.27.

It appears as Ch 1 of the cited book, previewable at https://mitpress.mit.edu/books/perception-beyond-inference

Koenderink's Vision as a user interface, a 2011 conference keynote paper , covers the same ground - it's downloadable at: https://www.researchgate.net/publication/253097219_Vision_as_a_User_Interface

Might I also suggest visiting http://www.gestaltrevision.be/en/ ,website of the GestaltreVision project, of which Koenderink is a co-principal investigator. The project's stated purpose is "to study Gestalt phenomena such as perceptual grouping and figure-ground organization using modern methods and techniques, and we want to integrate the Gestalt insights in our current understanding of how the visual brain works." 

Head for the Publications tab! 

Article Vision as a User Interface

Richard

re

http://journal.frontiersin.org/researchtopic/1713/radical-embodied-cognitive-neuroscience#articles

woohoo! what a goldmine! Thanks! Are you familair with Anthony Chemero's book with a very similar title?

Simon,

No, I confess Chemero's name is new to me. Thank you for what's turned out to be a priceless lead. From the synopsis of his book on https://mitpress.mit.edu/books/radical-embodied-cognitive-science , it's clearly another must-read for me. Equally so for many of his papers listed on https://uc.academia.edu/TonyChemero

As it happens, the first article of that list that I looked at was published within another Frontiers Research Topic under its Theoretical and Philosophical Psychology journal:

http://journal.frontiersin.org/researchtopic/1239/music-and-the-embodied-mind-a-jam-session-for-theorists-on-musical-improvisation-instrumental-self-e#articles

The topic's articles look to be sheer grist-to-the-mill for my own program. Perhaps you'll find them stimulating from your standpoint, too?

One goldmine leads to another. QED! I'm indebted to you.

Dear all:

I have been updating class materials for my graduate seminar "Art in the Embodied Mind," and I recently revised, expanded and updated the bibliography (actually more of a thematic menu than a reading list). It may be of interest to followers of this question, as it includes sections on visual perception, eye movements, and face recognition/evaluation, among others. Yesterday I uploaded the new version:

https://www.researchgate.net/publication/289251031_Art_in_the_embodied_mind_a_bibliography_updated_4_January_2016

Data Art in the embodied mind: a bibliography (updated 4 January 2016)

Dear Simon,

People have always taken the most sophisticated machine as metaphor of the brain. In the past, we used the clock metaphor and the steam-engine metaphor; now, we use the computer metaphor; in the future, we will undoubtedly use yet another metaphor based on a thus far unknown machine. Not because any of these metaphors is “true” (what is true if the visual world is a grand illusion constructed by our brain?), but because we have nothing better to work with.

As for the computer metaphor, I distinguish between a narrow version and a broad version – the narrow version being “The digital computer is a model of the neural brain”, and the broad version (a.k.a. information-processing metaphor) being “Information processing by computers is a model of cognitive processing by the brain.” Unlike the narrow computer metaphor and previous metaphors, the broad computer metaphor does not refer to the hardware principle that the brain is a physical system but it refers to software principles implemented in the brain to allow for cognition. My experience is that the latter is fruitful in investigating both cognitive and neuronal aspects of vision – but one indeed has to be clear about the definition of “information”.

In Shannon's (1948) classical information theory, information indeed has nothing to do with semantic content. He built on work by Nyquist (1924) and Hartley (1928), who defined the amount of information obtained from an event by the amount of uncertainty removed by the event, and who quantified this amount by the negative logarithm of the probability of the event. For instance, such an event could be the transmission of a message and a more likely message thus carries less information.

In the 1960's, a rethinking by people such as Kolmogorov and Solomonoff in mathematics and by people such as Garner and Leeuwenberg in visual perception led to modern information theory. Here, the amount of information carried by a “message” (say, a mental representation) is defined by the shortest description of the content of the message. Such short descriptions are obtained by capturing regularities, and a message that contains more regularity thus carries less information.

The latter definition of (amounts of) information differs fundamentally from Shannon's and is closer to the common sense meaning of the word “information” (also, e.g., in computer science). Without denying the merits of Shannon's approach, the latter definition works for me as an analogy of the way in which the brain might encode “messages” in an efficient and parsimonious fashion (near-optimal in Shannon's sense, but without loosing their content).

I addressed these issues (information and metaphors, respectively) in:

– van der Helm, P. A. (2000). Simplicity versus likelihood in visual perception: From surprisals to precisals. Psychological Bulletin, 126, 770—800. doi:10.1037//0033-2909.126.5.770. https://www.researchgate.net/publication/12332757_Simplicity_versus_likelihood_in_visual_perception_From_surprisals_to_precisals

– van der Helm, P. A. (2012). Cognitive architecture of perceptual organization: From neurons to gnosons. Cognitive Processing, 13, 13—40. doi:10.1007/s10339-011-0425-9. https://www.researchgate.net/publication/51800055_Cognitive_architecture_of_perceptual_organization_From_neurons_to_gnosons

Updated versions of these papers can be found in my book:

– van der Helm, P. A. (2014). Simplicity in vision: A multidisciplinary account of perceptual organization. Cambridge, UK: Cambridge University Press. http://www.cambridge.org/nl/academic/subjects/psychology/cognition/simplicity-vision-multidisciplinary-account-perceptual-organization?format=HB&isbn=9781107034341

Hope this is helpful.

Regards,

Peter

Article Simplicity versus likelihood in visual perception: From surp...

Article Cognitive architecture of perceptual organization: From neur...

Hello Peter

thankyou for your post.

> "People have always taken the most sophisticated machine as metaphor of the brain. In the past, we used the clock metaphor and the steam-engine metaphor; now, we use the computer metaphor;"

Yes, this argument was first made (afaik) by Jay Bolter in his pioneering book Turing's Man. He called it 'paradigmatic technolology' a term I often cite.

> "  we have nothing better to work with."

I'm not sure that assertion is true

> "As for the computer metaphor, I distinguish between a narrow version and a broad version – the narrow version being “The digital computer is a model of the neural brain”, and the broad version (a.k.a. information-processing metaphor) being “Information processing by computers is a model of cognitive processing by the brain.”"

I understand this distinction, it is good. But the second still depends on a whole set of assumptions about representational tokens in the brain - assumptions I do not feel are (yet) justified.

> "In Shannon's (1948) classical information theory, information indeed has nothing to do with semantic content. "

yes, as previousy noted.

>"In the 1960's, a rethinking by people such as Kolmogorov and Solomonoff in mathematics and by people such as Garner and Leeuwenberg in visual perception led to modern information theory. Here, the amount of information carried by a “message” (say, a mental representation) is defined by the shortest description of the content of the message. Such short descriptions are obtained by capturing regularities, and a message that contains more regularity thus carries less information.

The latter definition of (amounts of) information differs fundamentally from Shannon's and is closer to the common sense meaning of the word “information” (also, e.g., in computer science). Without denying the merits of Shannon's approach, the latter definition works for me as an analogy of the way in which the brain might encode “messages” in an efficient and parsimonious fashion (near-optimal in Shannon's sense, but without loosing their content)."

Thankyou for this later history, which I did not know.

I am no luddite, but talk of "common sense meaning of the word “information”" and "the brain might encode “messages”" makes me very leery. I remain dubious of the relevance of these information theoretic definitions of 'information' wrt the brain/body/self, and skeptical of what kind of ontological baggage words like 'messages' carry with them. Not opposed, but skeptical. I do not have a counter formulation, just a deep sense that the congitivist/computationalist paradigm creates thinking structures which actively impede understanding.

Simon, you're absolutely right to be critical and to question existing research paradigms. The problem with any new paradigm is that, at first, it may give a sense of deeper insight, but later, starts to wear off until it indeed may impede further progress. Yet, in the absence of a better paradigm, one often keeps on working with it. This is troublesome especially if its paradigmatic ideas are mistaken for the “truth” (whatever that may be). By the way, as I argued in my CogProc paper, I do not think that the cognitivist/computationalist paradigm has worn off, but I do think that it needs to connect to complementary concepts and ideas from, e.g., connectionism, dynamic systems theory, and neuroscience.

You're also absolutely right that the terminology in any paradigm carries a lot of questionable ontological bagage. This is nagging but I think also inevitable in our continuing quest for appropriate analogies, methaphors, and models by which we can only approximate the “real thing” (whatever that may be). In this sense, I am open to what is called a metaphysical (or ontological) reading of pluralism (which assumes that a "grand unifying theory" is possible), but for the moment, I adopt an explanatory (or epistemological) reading of pluralism – which, more pragmatically and in the spirit of David Marr, focuses on differences and parallels between existing explanations at different levels of description to see if and how they might be combined. My hope is that, eventually, this will lead to new and better thinking structures.

– Peter

Richard

> "I believe (i.e,, assume axiomatically) that meaningful discourse, even of the most primitive kind, is impossible without a foundation of axioms,"

I think this hits the nail on the head :) May I quote you?

On the other hand, is it true? There are kinds of knowledge that do not rely on axiomatic assumptions, as when I hit my head. Or when you hear middle C. In a paper I keep coming back to, Lakoff and Gallese (Where do concepts come from?) make a compelling argument for metaphors arising from embodied experience (as did Mark Johnson before them).

> "However, in your question's elaboration, your concerns about axioms and their confounding consequences for reasoning "veridically" about anything are stated most clearly. Your desire to abolish existing axioms leaves me, I admit, feeling it's unfair of me to request you to define "information", for, without you resorting to some axiomatic foundation upon which to formulate such a definition, surely, am I not asking the impossible?"

It seems to me that any system of 'knowledge' which is built on untested and untestable beliefs is fundamentally unscientific. Or at least, is rendered scientifically dubious. Which of course implies that (all?) science is religion, in the sense that it rests on 'faith'.

( I apologise to all philosophers for paddling in the shallows of scientific ontology - no doubt this is the kind of thing dealt with in scientific ontology 101. A little guidance would be appreciated ;)

There are of course other kinds of knowledge - that the sun will rise tomorrow, that my son loves me - which do not depend on scientific epistemology and these seem to have served humanity acceptably well for millennia, though there may be some debate on that subject.

As for a definition of information - it seems any definition of such an abstract and technical idea, as you say is, is impossible without relying on axiomatic assumptions, and 'its axioms all the way down' as they say. The very definition of bootstrapping.

Peter

>" You're also absolutely right that the terminology in any paradigm carries a lot of questionable ontological bagage. This is nagging but I think also inevitable in our continuing quest for appropriate analogies, methaphors, and models by which we can only approximate the “real thing” (whatever that may be)."

One of my strong misgivings in technical education is  about the slippage that occurs between teacher and students, and across generations, in the formalisation of 'disciplines'  in which subtlties and qualifications are lost in dogmatic simplification, resulting in mistaking the map for the only implicitly seen territory. The dogmatism of (neo)Cartersianism in comparison to Descartes own more inflected and subtle discussion of the mind/body issue is a prime example.

> "In this sense, I am open to what is called a metaphysical (or ontological) reading of pluralism (which assumes that a "grand unifying theory" is possible), but for the moment, I adopt an explanatory (or epistemological) reading of pluralism – which, more pragmatically and in the spirit of David Marr, focuses on differences and parallels between existing explanations at different levels of description to see if and how they might be combined. My hope is that, eventually, this will lead to new and better thinking structures."

Thankyou for such a succinct summary which is simultaneously philosophically sophisticated and pragmatic!

When mechanical automata were the rage, then the brain was likened to a clockwork machine. When hydraulics became a thing, then hydraulic models were described. When electronics rose in popularity, the brain was like electronics. Now the brain is like a computer. 

On the other hand, people say artificial intelligence is closer than we think...

Having eyes and yet not really seeing...no, "the eyes don't take in visual information"

"The only thing worse than being blind is having sight but no vision"- Helen Keller

Sight does not guarantee the formation of an visual image. Visual information is the product of the brain not the eyes (or the more philosophical notion of a heroic "soul" envisioning something). Hence vision is something apart from sight. The sighting of a painting versus the artistic vision that produced that painting. As attested by the popular expression,"hidden in plain sight".

Vision lies in the realm of attention and sight is mainly the product of mechanisms for projecting light sources or reflected light. More sublime is the connection between covert and overt attention shifts. Where the eyes may even "unlock" memories stored in the brain. Self-driving automobiles lack vision and not sight. Deriving and adapting a metaphor from the techno acronym: LIDAR: LIght (Eyes) +raDar (brain)

https://homepages.uni-tuebingen.de/klinikum/bianca.de-haan/deHaan_BrainRes_2008.pdf

Article Visual localization within LIDAR maps for automated urban driving