Thursday 30 June 2016

Taking Advantage Of Our Mistakes


‘Just to see what happens…’ is the sense of the word ‘experiment’ with which we are able to conceive of an experimenter who is lurching forward through more or less random behaviours in a limbo of ignorance, but with some (usually unjustified) air of optimism. It is also the sense in which, in the course of doing something that the experimenters do know how to do (such as boiling flasks of urine), they unexpectedly discover how to do something that they did not know how to do and thereafter—if an unexpected but desirably efficacious outcome has emerged—are able to do ‘the same thing’ again, although now in a differently purposeful way. (Donald Brook 2015)

My 5 year old son recently mistook a photograph behind a shop counter for an actual person. After a brief moment of surprise, he remarked: “That could scare other children.” I had noticed the picture too, but hadn’t been fooled by it, partly I think because my vantage point made me less prone to the illusion. My son’s comment seemed to be fairly trivial at the time, but it left me wondering about the nature of mistakes. Most especially it left me thinking about how discoveries and insights are often the unexpected offspring of mistakes.

“A discovery is said to be an accident meeting a prepared mind.” -Albert Szent-Gyorgyi

Invariably when we make discoveries, they arise unexpectedly. You cannot get up in the morning and decide to have a discovery, an epiphany, a revelation or an insight. It is in the nature of these concepts, that their treasures are stumbled upon “in the limbo of ignorance”, even when our anticipatory bags are diligently packed in preparation. No matter how well prepared we might be, no matter how well equipped, we can only fully anticipate what we already know or can imagine. There would be nothing to discover, no insights to be had and no revelations to befall us if we were already acquainted with our discoveries in advance.

During a tutorial a few years ago, I found myself saying: “A mistake is a discovery never to be repeated.” I made a point of committing this phrase to memory for future use, but I’m beginning to realise that it isn’t the pithy truism that I had assumed. Sometimes there is potential in our tendency to make mistakes, especially mistakes of a commonly occurring and therefore exploitable sort. Indeed, there are times when our mistakes become what James Joyce called: “portals of discovery”.

When my son mistook the shop image for an actual person, he made a fairly commonplace mistake. But when he remarked that the image could scare other children, he had already transformed his mistake into an insightful observation: other people like him would have a similar response in similar circumstances. In essence he had discovered, or at least he was noting the fact, that in certain circumstances images can be mistaken for the things they represent.

In my investigations of the theories of Donald Brook, I have often been struck by the emphasis he places upon the notion of “sensory discrimination failure”. It has often seemed to me that he places too much emphasis on what we fail to perceive and not enough on what we succeed in perceiving. In my attempts to explore and exploit his insights I have frequently avoided the notion of failure, preferring instead to draw attention to the “characteristics” of our sensory system or our sensory “limitations”. But if it is true that our perceptual mistakes can be the source of insight—of discoveries of how we and others like us are susceptible in the same purposefully exploitable ways—then perhaps my uneasiness has been unwarranted. To be fair to Brook, on several occasions he has mentioned that sensory discrimination failure is a great “felicity”. After all, amongst other things, it enables us to successfully simulate three-dimensional objects with the flat things we call “pictures” or—by a completely different method of substitution—to imitate (with various levels of success of course) the sounds of innumerable different creatures or to replicate, copy, mimic and emulate all manner of things.

Non-verbal matching and non-verbal simulating are two significantly different ways of exploiting the useful substitutability of one thing for another thing under certain circumstances, for certain communities of perceivers, for certain purposes, by virtue of the fact that all sensory systems are systematically unreliable in certain ways. Science is the most orderly and progressive way of finding out what we might be expected to say about the world if our sensory systems were not unreliable. (In spite of this--and not paradoxically--we could never have got language, and hence science, going at all if our sensory systems hadn't been unreliable in the felicitous way they are, enabling us to substitute one thing for another thing in ways that have an evolutionary pay-off). (Brook, in personal communication 27/12/12)

That perception is usually successful is obvious. No doubt this is why Brook’s primary concern has always been to elucidate the contrasting importance of perceptual mistakes in our account of representation and perception. Unlike mistakes of a more procedural sort, perceptual mistakes cannot be overcome by careful practice or training. We can be more or less vigilant of course, but illustrators, photographers and even perceptual psychologists are no less susceptible to well presented illusions than the rest of us. There is no sense in which my son could have avoided his mistake by trying a little harder or by being better informed. Such illusions do not result from a lack of knowledge or expertise, they result from our creaturely fallibilities. Where our skills come to the fore is in the dawning realisation that we have made a mistake, not in the mistaking. And by the time we have figured out what is going on, the mistake has already been made.

Some people take the view that we would have language even if we had no inability to discriminate between certain things in certain circumstances. But what would this entail? To be able to discriminate between everything and anything under all circumstances and in all respects would be to exist in a universe with no regularities and patterns whatsoever. Nothing would be the same as anything else and even similarity would be out of the question because there would be no end to our discriminations. In such a universe there would be no possibility of substituting or exchanging one thing for another equivalent thing because no two things would be equivalent. Language simply couldn’t get the slightest foothold in such a universe because nothing would be like anything else at all. It is hard to imagine a more alien universe or one more unsuited to communication. To paraphrase Terry Eagleton’s rendition of Wittgenstein: “It makes no sense to speak of perceiving something in a context where we could not possibly make mistakes."

Thursday 23 June 2016

Can Information Be Naturalised?



Psychologists Sabrina Golonka and Andrew Wilson have recently shared a yet to be published paper entitled: “Ecological Representations”. I noted their work in the summary literature review I posted here a couple of weeks ago, but from this new paper it would appear that that they have shifted their position on the question of cognitive representation considerably. They write: “We will agree that cognition requires representations.” Hopefully they can be persuaded that this is only true if the required representations are of the fully public and intentional sort and not the neural and non-intentional sort that they seem to have embraced.

The influential psychologist J. J. Gibson, is well known for his rejection of representationalism. His work on perception is foundational to many of the ideas pursued by Golonka and Wilson. At the core of G&W’s argument is the conjecture that “Gibson’s ecological information fits the basic definition of representation.” They observe that most “radical” theories of embodied cognition are based on Gibson’s ecological approach to perception and action, and that, despite some successes, these theories have not made significant headway in explaining higher order cognitive processes such as thinking about absent objects etc. They claim to have discovered a way to salvage the good work on all sides of the debate. I aim to show that their proposed solution comes at an unacceptably high price.

In Chapter 8 of Gibson’s book “The Ecological Approach to Visual Perception”, (1979) Gibson coins the term “affordances” to describe what he suggests the environment “offers” animals for their survival. He writes:
[I]f there is information in light for the perception of surfaces, is there information for the perception of what they afford? Perhaps the composition and layout of surfaces constitute what they afford. If so, to perceive them is to perceive what they afford. This is a radical hypothesis, for it implies that the “values” and “meanings” of things in the environment can be directly perceived. Moreover, it would explain the sense in which values and meanings are external to the perceiver.
The hypothesis that values and meaning are external to perceivers corresponds closely with the philosophy of Ludwig Wittgenstein, who, towards the end of his life, developed some carefully nuanced arguments to show how values and meanings are best understood as socially negotiated and rule dependent practices rather than inner states of perceivers. However, unlike Gibson, Wittgenstein almost certainly would not agree that we actually perceive meanings and values, whether directly or otherwise. To put the point as simply as possible, the value of money is not a perceptible property of the coin or note in your pocket. Value is ascribed to things by virtue of practices of exchange that involve the treatment of things as if they have properties that they do not in fact possess. Indeed, without the capacity to pretend and to accept acts of pretence, the skills necessary to ascribe value and meaning to things would be out of the question.

So when Gibson proposes that we perceive the affordances the “environment…offers… provides or furnishes”, he confuses practices of use attribution and/or meaning ascription with skills of perception. I think this is a very serious mistake that Golonka and Wilson only amplify with their new paper. Wittgenstein took the view that the meaning of a word is best determined by looking at the various ways in which it is used. The Scottish Enlightenment Philosopher, Thomas Reid made a very similar point about 300 years ago. On several occasions Wittgenstein also suggested that we should regard words as tools. Do tools have perceptible affordances? According to Martin Heidegger a tool has a "usability" that "belongs to it essentially". Wittgenstein would disagree. Do we perceive the use of a tool we have never seen before? Think of a fork. Would we immediately see its alleged inherent Heideggerian function if we were intelligent animals of a different shape and size? G&W are bound by the force of reason to say that we do not. So then, how can the many different ways of using a stick  its alleged affordances  be perceptible in the stick?

Another serious issue that arises in Gibson’s theorisation, and that G&W further ramify, is his suggestion that light carries information; that there is information in it (I will return to this issue of “content” in a moment). The philosophy of information (as distinct from Information Theory which is an engineering term) was in its infancy in Gibson’s day (some say that it still is (Floridi 2011)), so it is unlikely that Gibson would have been aware of the dangers of his use of the word. “Information” is what Ryle (1954) might have called a “smother word”. For Ryle, terms like “depiction”, “description” and “illustration” often smother important conceptual distinctions and create otherwise avoidable philosophical dilemmas. It is the task of conceptual analysis to tease out these differences and to dispel conceptual confusion.

I have mentioned before on this blog how even the early Wittgenstein misleadingly described language as a "picture" of the world. I have also discussed how C.S Peirce regarded the whole universe as being perfused with meaningful signs. Grice (1957) too, saw no confusion is assuming that nature creates “natural meanings” in addition to the “non-natural” ones that we humans generate. More recently, Fred Adams published a paper (2003) attempting to “naturalise meaning” and to suggest a way to account for the meaningful content that he believes is realised in the mind/brain. He writes: “To be of value to a would-be knower, or to someone interested in naturalizing the mind, information must be an objective, mind-independent commodity.” He provides the following two examples as evidence of this supposed natural commodity (if that isn’t already a careless oxymoron):

Waves of radiation traveling through space may contain information about the Big Bang before anyone detects it. Fingerprints on the gun may contain information about who pulled the trigger before anyone lifts the prints. Thus, information appears to be mind-independent (and, thereby, language independent too).

According to a recent comment from Golonka on their blog, they “agree with content critiques regarding mental reps”, so they would probably reject at least some of Adams’ radical representationalism. Nonetheless, since they take Gibson’s ecological information to fit with ecological representations they have a job on their hands to reconcile their agreement with say Hutto and Myin (2013) on the question of content and their own representational “vehicles”. If, as I contend, the influence is merely causal, then no representation, no vehicles and no content need be imputed.

G&W are clearly aware that perhaps the greatest explanatory challenge for a theory of cognition is to give a coherent account of intentionality. In philosophy "intentionality" has a technical sense that I assume is the sense in which G&W are using it. Nonetheless, both senses are applicable here. They state that: “The need for intentionality therefore provided the first and primary motivation for treating cognition as necessarily representational.” What should be pointed out here is that this assumption is questionable on grounds of logical incoherence. In order for cognition to be intentional (in either sense), it must be intended, but if it is intended this intention must (according to the logic of the argument) be supplied by representations, then these must also be intended and must therefore be motivated by further intentionally generated representations. This is a logical regress of the most vicious kind that is widely overlooked in much of the relevant literature. Perhaps it is this general lack of recognition that has led to G&W's oversight of this serious logical obstacle.

G&W do acknowledge the “symbol grounding problem” though. This is characterised as the challenge of explaining how symbols gain their meaning (their representational content in fact) outwith a system of mutually agreed rules. This is another serious challenge to representationalism that, for example, Adams fails to mention at all. He evidently takes it as unchallenging that fingerprints “contain” information. 

Words like “contain” and “content” are a common cause of conceptual confusion. When we talk of the “content” of a painting, we do not mean that the content is a property analogous to the size and shape of the painting. Content is not a special characteristic of objects. It is not perceptible. If anything, content is a special characteristic of us; of the things we can do, not something that inheres in things ready to be extracted like some kind of magical inform-essence. Fingerprints are part of a forensic system. They are meaningless outwith this system. Our knowledge imbues nature with meaning but in the process it leaves nature entirely untouched, in respect of its content that is.

G&W also wisely acknowledge the “system-detectable error problem” (Bickhard, 2009). Within any notionally intelligent system there has to be a way for the system to detect and avoid errors. Once again, within a social system this process depends on the observance of various socially negotiated rules. But without such rules it is challenging to say the least, to know how errors could even qualify as errors, let alone be avoided. Like most of the obstacles to representationalism, the issue here comes down once again to intentionality. In order to detect errors you need a system that can represent and compare errors with successes and in order for the system to represent the difference between error and success, evaluative criteria or some form of metric is needed by which such comparisons can be made.

In their definition of representation, G&W begin, rightly, by stating that representations are stand-ins. However they then rely heavily on Newell (1980) who was principally concerned with symbol systems and “designation”. Newell defines representation/designation thus:

An entity X designates an entity Y relative to a process P, if, when P takes X as input, its behavior depends on Y.

In my view this is too narrow a definition of representation. Onomatopoeia does not designate the thing it represents and nor does a photograph, an enactment or a model. Designation is more akin to delegation, nomination or stipulation than it is to depiction or imitation. So, at best, Newell’s definition applies to symbolic representations only. However, to be fair to G&W they do a quite good job of translating Newell’s formulation into a more palatable version:

X, is a thing that is not Y but can close the gap and that P can access and use as if it were Y; when it does, P works as if it had access to Y.

This can be tidied up as:

X is a thing that P can use as if it were Y. When it does, P works as if it had access to Y.

So, on this basis:

A wind turbine is a thing that a lightbulb can use as if it were a battery. When it does, the lightbulb works as if it had access to a battery.

Or, better still:

Sugar syrup is a thing that a honeybee can use as if it were honey. When it does, the honeybee works as if it had access to honey.

I may be missing something important, but I fail to see how this qualifies the wind turbine as a representation of a battery or sugar syrup as a representation of honey. All of the paradigmatic cases of representation of which I am aware involve substitution for the purposes of communication between agents, not simple replacement of functional component A with alternative functional component B. The radioactive isotope of strontium substitutes for calcium in bone formation but it certainly isn’t a representation of calcium. Something is awry in Newell’s formulation.

Leaving this objection aside for the moment, G&W focus their attention on what they see as “the gap” which X can “close” between P and Y (the bee and its honey). But this is merely an anomalous consequence of their turn of phrase (which I edited out of my reformulation). Sugar syrup does not close a gap between the bee and its honey; it simply replaces honey. Nonetheless G&W spend several sentences fleshing out the significance of this supposed “action at a distance”.

G&W turn next to a consideration of “ecological information [as] a representation”. They define ecological information as energy patterns of  “lawful interaction of the energy with the dynamics of the world [that] are used by organisms to perceive that world.” [My emphasis]. If organisms use energy patterns to perceive the world, then this form of usage needs to be sharply distinguished from intentional use, otherwise we have no means of distinguishing tool using creatures (humans mostly) from all the other creatures in the world who do not use tools. Moreover, we also need this important distinction to distinguish between the intentional actions of purposeful creatures and the efficacious (but not intentionally directed) behaviours of their internal processes. My bone forming processes do not intend to use strontium as a replacement for calcium, but my dentist did intend to use gold as a crown for one of my teeth. This is why my crown is plausibly a representation—indeed it is a cast—of parts of the tooth it replaced. The reason such actions, as the replacement of a tooth, are intentional is because they are performed in pursuit of a goal that can be represented on demand. The fact that my dentist could explain his behaviour is not because a representation of my tooth was contained in his neural fibres but because the capacity to represent the aims of his activity was something he could do; something he could perform as a competent agent embedded in a culture where such actions are understood.

If I might be allowed to go into a little technical detail, theorists often distinguish between teleological and teleonomic descriptions of behaviour. A telos is a goal, an aim or an envisaged end that an action is intentionally directed towards. Teleological behaviours are thus genuinely purposeful actions. Teleonomic behaviours, on the other hand, often have the appearance of purposefulness but are actually merely efficacious (some theorists use the word “purposive” here, as contrasted with genuinely purposeful activity), having been shaped by millions of years of evolution. When we say that a plant uses varying light intensities to find its way towards the sun, we do not mean to suggest that the plant is an intentionally directed agent: a perceiver. We are simply using a teleonomic description. Unfortunately I think both Gibson and G&W conflate teleonomic descriptions in which organisms and their inner processes “use energy and genuinely teleological descriptions in which we human agents use energy—to illuminate a light bulb for instance.

I do not believe that perceivers use energy in the way that both Gibson and G&W suggest. I might use my desk light in order to read a book at night, but the inner processes that in large part bring about my perception of the book do not use either the desk light or the energy patterns that emanate from it in this intentionally directed way at all. I can choose to turn out the light, but my inner processes have no choice in the matter. Choices are exercised by whole agents, not by their parts (Hacker and Bennett 2007).

A lot of confusion can be cleared up in discussions of representation if we distinguish sharply between processes in which X is taken as Y and actions in which X is treated as Y. My bones will take strontium as calcium but only a performer of actions can treat an act of mock aggression as if it is merely playful as opposed to genuinely threatening. This is why I argue that pretending is the most fundamental and important skill in intelligent behaviour because it is the basis of the higher forms of cognition that G&W are so keen to account for.

G&W return to the notion of “a gap” when they state: “Most of the behaviorally relevant dynamics in the world are ‘over there’ and not in mechanical contact with the organism. They must therefore be perceived.” The fact that the keys of my keyboard are “over there” and not in “mechanical contact” with my fingers does not mean that they are not causally influential upon me by virtue of the light reflected from them. My perception certainly depends upon light but my perception is not of the light as information, it is of the keys as keys. Light is something we know about, not something we see. So, whilst it is true that we pretenders can act as if light is perceptible, the light reflected from my keyboard is simply taken by my sensory system not as information but as causal influence. When G&W say that: “Perception relies on information about dynamics” this is not true. Only knowledge (propositional knowledge that is) relies on information about dynamics.

According to G&W:

Gibsonian ecological information is only a kinematic projection of those dynamics into an energy array. […] This means that kinematic information cannot be identical to the dynamical world, and this fact is effectively a poverty of stimulus.

Kinematic information is quite clearly a culturally enabled ascription—indeed a “description”—of “units” of measure to the “dynamical world”. There is no possibility that such sophisticated cultural contrivances as units are to be found in nature.

Their worries about “a gap”, “action at a distance” and “a poverty of stimulus” continue when they write:

Other lines of neuroscientific enquiry do suggest that at least some of the structure of energy impinging on perceptual receptors is preserved as it travels through the nervous system.
According to G&W’s theory of representation it is important that structure is carried through the nervous system because this qualifies the structure as a neural representation of the ecological information that caused it (recall that they take all forms of replacement to be representational). At the risk of repeating myself, the fact that some pattern corresponds with an antecedent state of affairs does not mean that the pattern is a representation. Effects are not representations of their causes. If they were, then the universe would be nothing but representations. I therefore think we have good reason to reject G&W’s proposal that “at least some of the neural activity caused by informational representations will qualify as a neural representation of that information.”

To be fair to G&W, they observe that: “These neural representations are… not implementing the mental representations of the standard cognitive approach.” because they do not “enrich, model or predict anything about that information.” If this is true, then it leaves these representations as representations in name only.

Later in their paper G&W attempt to tackle the issue of higher order cognition. They remark: “To be clear, the stipulation that knowledge systems must be conceptual and componential is so that knowledge systems can support counterfactual thinking, etc.” This is mistaken. Pretending that I am rocking a baby in my arms is a gesture that would be understood by humans the world over but, even though it is counterfactual (there is no baby after all) it is not a conceptual representation. Conceptualisation relies on the ability to manipulate abstractions and there is no other species on the planet that has the capacity to manipulate abstractions with anything more the most rudimentary competence.
Washoe, the first of the signing apes, had been regularly bathed. Sometimes between the ages of one and a half and two years, she picked up her doll, filled the bathtub with water, dumped the doll in the tub, took it out and dried it with a towel. In later repetitions she even soaped the doll. This is imitation, but it also must be a form of representation—indeed, of pretence. (Jolly 2000, 291)
On page 18 G&W write: “From the first person perspective of the organism, it is just interacting with information.” We commonly interact with others by means of information but it is somewhat confused to suggest without qualification that we interact with information. Our use of information forms part of our interactions with other intelligent agents: people usually. When we use so called “interactive technologies” we do so in a sense that is derivative of these interactions with other agents. Information is simply not responsive in the way that that other intelligent agents are. It helps to regard information as a tool. We use our tools but it is somewhat strained to say that we interact with them.

In conclusion, G&W are right to regard representation as important in the explanation of higher order capacities but only if we regard representation as a thoroughly public activity of intelligent agents. G&W are also right to focus on behaviour that treats X as if it is Y. Nonetheless their Newell-derived definition of representation is inadequate to the task of distinguishing between behaviours in which X is taken for Y and actions in which X is treated as Y. If they were to thoroughly examine this important distinction, they would probably recognise that representation is the point of demarcation between evolved efficacious processes and behaviours and de facto teleological actions; between nature and culture. Information cannot be naturalised because information is a cultural contrivance.




Thursday 16 June 2016

Dretske’s Dreadful Theory Of What We See



In a presentation from 2008, the late (as of 2013) Stanford philosophy professor Fred Dretske argues that seeing includes the perception of objects (including relations between objects), properties (shape, size and colour etc.) and facts. For Dretske, the facts we perceive are “things we come to know by seeing”. He claims that there is a danger that we take a failure to notice or detect objects and properties as a failure to actually see them. He presents the following two images in order to explain this claim.
Image A
Image B

Image B contains an additional shape. In his presentation, Dretske presents the images one after the other and inserts a transition between the two that makes it difficult to see the difference between them. In Dretske’s view: “One should not conclude from the fact that you didn’t see that there was a difference to the conclusion that you didn’t see the object that made the difference.” He remarks:
Even if you don’t detect it, even if you don’t notice it, even if you don’t know that you are seeing different things. Even if you don’t see the fact that there is a difference, you still might see the objects and properties that make the difference. You just don’t realise you do.
It should be obvious that if you do not detect, realise or notice that the traffic lights are red, then you cannot be said to see them (at least in respect of their being red). Nonetheless Dretske holds that seeing is independent of conscious awareness. In his view, we can genuinely see objects and properties without detecting, noticing or knowing that we do so. Dretske uses this questionable conclusion to promote his theory of conscious experience. He claims that: “Your experience of an object is conscious if it gives you knowledge of that object.”
For Dretske, all knowledge is knowledge of facts, and in order for an experience to be conscious, it must provide such knowledge. But this is absurd. If we look at a familiar object, we do not lose consciousness of it because it ceases to give us knowledge. If consciousness depends on the acquisition of knowledge, then all lapses in attention must be accompanied by lapses in consciousness. This is quite evidently not the case.
We perceive the phonemes, morphemes and sentences in which facts are typically stated but it is debatable whether we actually perceive facts at all. Earlier today I asked my partner “When did you last see a fact.” She looked at me quizzically and remarked: “You don’t see facts.” Some people might contend that texts, graphs and diagrams etc. can be seen and thus are visible facts, but it is important to note that representations of facts are not the facts they are used to represent. Many facts can be demonstrated, but it does not follow that a demonstration of a fact can be reduced to the fact it is intended to demonstrate.
Earlier in his presentation, Dretske states: “The facts we see—the things we come to know by seeing—we come to know them by seeing objects and their properties.” Clearly then, Dretske realises that perception of things is not the same as the “perception” (his use, not mine) of facts. Facts must be derived in some way from perception, but quite how they are derived Dretske neglects to mention.
The absurdity of Dretske’s position becomes more pronounced towards the end of his presentation. He argues that when we look at a wall of 350 bricks for a few seconds, we acquire 350 “distinct pieces of knowledge—one for each brick in the wall.”  Not only this, but during the Q&A session it becomes clear that he thinks we acquire “infinite” knowledge “for free” when we look at things. The reason he believes this is because he is committed to the idea that the perception of facts involves the acquisition of what he calls “tacit knowledge”.
You don’t have to actively think something to know it. There are a great many things you know tacitly, not because when you acquired the knowledge you are actually thinking about it or believing it but because you have the kind of experience of it which, if you do later think about it, will tell you what you need to know.
There is no such thing as a memory “telling” us anything, so Dretske can only mean that we have to interpret an experience ("later think about it") to derive factual knowledge from it. If we do not think about it—if we do not interpret the memory by means of our conceptually enabled inferential skills—then we cannot be said to have yet formed any factual knowledge on the basis of the memory. An analogy will help. If we have a workshop full of materials, we may be capable of using these to create some tools, but unless we actually manufacture these tools, we cannot claim to possess them. Tacit knowledge, and knowledge in general in fact, is not something that we possess like a scar or a souvenir, it is something we are capable of doing, something we are capable of bringing about by the application of skill.
When we fail to see a difference between two different things, at least two factors need to be taken into consideration. Firstly, our sensory organs may be limited by evolutionary constraints that give rise to regularly occurring fallibilities in certain circumstances. Secondly, the circumstances of encounter (the illumination, position, angle of view, delay between images etc.) may limit discrimination more than might otherwise be the case. Certainly any actual differences within our visual field may have a causal influence upon our sensory system, but this does not mean that these influences are dealt with at some non-conscious, unconscious or “sub-personal” (Dennett 1987) level of seeing that we "just don't realise". It just means that there are various complex processes involved in perception that are not conscious and that some of these have insufficient influence to rise to the level of purposeful action: consciousness.

At the 25:00 minute mark of the presentation, Dretske discusses an image of a square divided into nine coloured portions. He claims that if each of the nine squares was coloured the same shade of blue and placed so close together that we couldn't see the edges between them, we would still see nine squares. This is like arguing that we see the screen when we watch a movie. If pressed, we would probably agree that the screen is visible while we watch Star Wars or Gone With The Wind etc. But what we would be very unlikely to concede, and what follows from Dretske's thesis, is that we see each pixel on the screen and thus gain millions of distinct pieces of knowledge—one for each pixel. It should also be noted that pixels are themselves divisible into individual photons. Do we have distinct pieces of knowledge of photons too? 

At some point we have to acknowledge that all sensory systems are limited in various ways and it therefore follows that these limitations make it impossible to discriminate between things (that may in fact be quite different) in certain circumstances and in certain respects. It is by virtue of differences that all creatures discriminate between things. Without sensory discrimination there would be no life. Nonetheless, sensory discrimination is not free of limitations. Without these limitations, there would be no question of our mistaking any one thing for another different thing, and there would also be no question of our accepting a flat thing as a viable stand-in of a three-dimensional thing. The fact that things of one sort (images say) can be mistaken for things of another sort (three dimensional objects for example) makes images supremely apt as representational tools; as things that can stand-in for the objects they represent.
I suggest that any explanation of perception that fails to account for the role of discrimination failure within our practices of nonverbal representation is probably doomed.