30 June 2011

Inferential Perception

Filed under: Psychology, Reflections — ktismatics @ 9:27 am

My morning walk took me on a mesa overlooking a cow pasture. In the corner of the pasture is a pond and, this morning, in the pond was a… what? The first feature I could distinguish was the head, lifting out of the water on a longish neck. The torso was spotted black and white. Four legs could be discerned. Not a cow, surely: too skinny for that. A horse? It had the triangular head for it. But there was something funny about the head: bright white, nearly gleaming; triangular like a horse’s, but if anything too triangular, like a poorly-drawn head of a horse, and seemingly too big for the body supporting it. And the legs: why could I see all four of them extended, as if the horse was floating on its side? But surely it wasn’t dead, since the first feature I had seen was the head, stretched up and out of the water on its long neck. My attention was momentarily distracted by cars moving along the road beyond the far edge of the pond. Looking back to the pond I suddenly realized that the horse was actually larger than the cars, maybe twenty feet long — far too big to be a horse. And the pattern of black and white spots had shifted, and was in fact still shifting as I continued to walk at an oblique angle to it. The pond, in contrast, retained its consistent flat grey-green color. Then, suddenly, I realized that I had it all wrong. What I’d been looking at wasn’t an object in the pond. It was a part of the pond’s surface, the only part not covered by pond scum. What I had taken to be the mottled pelt of an animal turned out to be a reflection of the partly cloudy sky behind it, its pattern changing with the wind and with my movement relative to the reflective patch of pond surface.

Smaller, more intricately patterned in light and shadow, more irregularly shaped than its background, this patch of clear pond carried all the visual signals by which one typically distinguishes figure from ground, object from context. Trying to discern what sort of thing this anomalous object actually was — during those two or three seconds of confusion I had engaged in an act of conscious attention and categorization and inference, whereby I repeatedly compared the features and the whole of what I was seeing with abstracted representations compiled in my memory from other specific objects I’d previously encountered and for which I had names. But the initial perception of this shape as a 3D object backgrounded against the 2D surface of the pond: that was nearly instantaneous, preconscious. But it wasn’t a direct perception. What hit my retinas as I looked at the pond was a 2D array of light in varying colors and luminances. Nearly instantaneously and unconsciously, I had transformed this luminance array into a perceptual representation of the 3D landscape I was looking at, assembled not only from the immediate sensory input extracted from the world but also from memory-driven unconscious inferences for making sense of the sensory input. It’s how perception works all the time.

*  *  *

In his clear and thorough and excellent post about concepts, Pete Wolfendale describes, via Kant and Brandom, the iterative process by which people revise their judgments about the world based on experience of the world. “This is,” says Pete, “the theoretical role of reason in constituting a unified account of nature.” Pete endorses what he calls “thick practices” of discerning reality:

“The causal features of the objects themselves act as constraints upon the development of our dispositions to respond to them in perception and action, and via them upon our dispositions to reason about them.”

Our “dispositions to respond” iteratively and inferentially to the world need not be stated explicitly, in the form of rules for understanding the world and for sorting its objects into categories. Usually the norms are implicit, to be inferred from patterns of  practice in engaging the world. Making norms explicit, in the form of statements communicated via language, is essential if humans are going to help one other recognize errors in their idiosyncratic subjective understandings of the world, and if they are going to learn new truths about the world discovered by others. But it’s also the case that an individual in isolation can correct an erroneous understanding of the world through activating the iterative practice of comparing features of the world with dispositions for interpreting those features.

While it’s possible to describe the iterative processes underlying perception propositionally, there is no reason to assert that the practice is intrinsically linguistic. Cognitive processes can operate without ever coming into conscious awareness. Conscious thinking is a relatively slow process, requiring focused attention on a relatively small number of features extracted from the world or from memory. In perceiving one’s immediate environment it’s more efficient, and more thorough, to rely on distributed, unconscious cognitive operations for iterating between sensory input and representations of prior perceptual arrays stored in memory. Perceptual inference need not rely on implicit reason or propositional logic, as evidenced for example by the observation that many other mammalian species improve through experience their abilities to navigate environments, not just by following a few well-defined trails but by staking out new paths and shortcuts they’ve never taken before.

The contention that inferential processes for understanding the world need not be conscious or even rational does not obviate Pete’s ideas about the importance of concepts and propositions in achieving a better understanding of what the world is really like. Though human reasoning and language are qualitatively different from and arguably better than the cognitive capacities of other animals, these abilities didn’t just come out of nowhere. The ability iteratively to compare concepts with the objects they represent evolved incrementally from similar comparative practices that don’t rely on concepts. And, like other mammalian species, we still deploy these non-conceptual iterative practices whenever we are actively perceiving the world we live in.



  1. “Making norms explicit, in the form of statements communicated via language, is essential if humans are going to help one other recognize errors in their idiosyncratic subjective understandings of the world, and if they are going to learn new truths about the world discovered by others.”

    Genesis 41

    1 When two full years had passed, Pharaoh had a dream: He was standing by the Nile, 2 when out of the river there came up seven cows, sleek and fat, and they grazed among the reeds. 3 After them, seven other cows, ugly and gaunt, came up out of the Nile and stood beside those on the riverbank. 4 And the cows that were ugly and gaunt ate up the seven sleek, fat cows. Then Pharaoh woke up.
    5 He fell asleep again and had a second dream: Seven heads of grain, healthy and good, were growing on a single stalk. 6 After them, seven other heads of grain sprouted—thin and scorched by the east wind. 7 The thin heads of grain swallowed up the seven healthy, full heads. Then Pharaoh woke up; it had been a dream.

    8 In the morning his mind was troubled, so he sent for all the magicians and wise men of Egypt. Pharaoh told them his dreams, but no one could interpret them for him.

    9 Then the chief cupbearer said to Pharaoh, “Today I am reminded of my shortcomings. 10 Pharaoh was once angry with his servants, and he imprisoned me and the chief baker in the house of the captain of the guard. 11 Each of us had a dream the same night, and each dream had a meaning of its own. 12 Now a young Hebrew was there with us, a servant of the captain of the guard. We told him our dreams, and he interpreted them for us, giving each man the interpretation of his dream. 13 And things turned out exactly as he interpreted them to us: I was restored to my position, and the other man was impaled.”

    14 So Pharaoh sent for Joseph, and he was quickly brought from the dungeon. When he had shaved and changed his clothes, he came before Pharaoh.

    15 Pharaoh said to Joseph, “I had a dream, and no one can interpret it. But I have heard it said of you that when you hear a dream you can interpret it.”

    16 “I cannot do it,” Joseph replied to Pharaoh, “but God will give Pharaoh the answer he desires.”

    17 Then Pharaoh said to Joseph, “In my dream I was standing on the bank of the Nile, 18 when out of the river there came up seven cows, fat and sleek, and they grazed among the reeds. 19 After them, seven other cows came up—scrawny and very ugly and lean. I had never seen such ugly cows in all the land of Egypt. 20 The lean, ugly cows ate up the seven fat cows that came up first. 21 But even after they ate them, no one could tell that they had done so; they looked just as ugly as before. Then I woke up.

    22 “In my dream I saw seven heads of grain, full and good, growing on a single stalk. 23 After them, seven other heads sprouted—withered and thin and scorched by the east wind. 24 The thin heads of grain swallowed up the seven good heads. I told this to the magicians, but none of them could explain it to me.”

    25 Then Joseph said to Pharaoh, “The dreams of Pharaoh are one and the same. God has revealed to Pharaoh what he is about to do. 26 The seven good cows are seven years, and the seven good heads of grain are seven years; it is one and the same dream. 27 The seven lean, ugly cows that came up afterward are seven years, and so are the seven worthless heads of grain scorched by the east wind: They are seven years of famine.

    28 “It is just as I said to Pharaoh: God has shown Pharaoh what he is about to do. 29 Seven years of great abundance are coming throughout the land of Egypt, 30 but seven years of famine will follow them. Then all the abundance in Egypt will be forgotten, and the famine will ravage the land. 31 The abundance in the land will not be remembered, because the famine that follows it will be so severe. 32 The reason the dream was given to Pharaoh in two forms is that the matter has been firmly decided by God, and God will do it soon.

    33 “And now let Pharaoh look for a discerning and wise man and put him in charge of the land of Egypt. 34 Let Pharaoh appoint commissioners over the land to take a fifth of the harvest of Egypt during the seven years of abundance. 35 They should collect all the food of these good years that are coming and store up the grain under the authority of Pharaoh, to be kept in the cities for food. 36 This food should be held in reserve for the country, to be used during the seven years of famine that will come upon Egypt, so that the country may not be ruined by the famine.”

    37 The plan seemed good to Pharaoh and to all his officials. 38 So Pharaoh asked them, “Can we find anyone like this man, one in whom is the spirit of God[a]?”

    39 Then Pharaoh said to Joseph, “Since God has made all this known to you, there is no one so discerning and wise as you. 40 You shall be in charge of my palace, and all my people are to submit to your orders. Only with respect to the throne will I be greater than you.”


    Comment by erdman31 — 30 June 2011 @ 11:00 am

  2. It’s interesting that in the Hebrew scriptures, dreams are significant for interpreting the future, or giving guidance and discernment. In Genesis 41, Joseph rises to prominence by decoding the dream, foretelling the future famine, and then subsequently making recommendations for sustaining the land. In Daniel 2, there is a similar story. (Personally, I thought the Genesis 41 story might be particularly interesting, given that your dream seemed to involve livestock!)

    Daniel 2

    1 In the second year of his reign, Nebuchadnezzar had dreams; his mind was troubled and he could not sleep. 2 So the king summoned the magicians, enchanters, sorcerers and astrologers[a] to tell him what he had dreamed. When they came in and stood before the king, 3 he said to them, “I have had a dream that troubles me and I want to know what it means.”
    4 Then the astrologers answered the king, “May the king live forever! Tell your servants the dream, and we will interpret it.”

    5 The king replied to the astrologers, “This is what I have firmly decided: If you do not tell me what my dream was and interpret it, I will have you cut into pieces and your houses turned into piles of rubble. 6 But if you tell me the dream and explain it, you will receive from me gifts and rewards and great honor. So tell me the dream and interpret it for me.”

    7 Once more they replied, “Let the king tell his servants the dream, and we will interpret it.”

    8 Then the king answered, “I am certain that you are trying to gain time, because you realize that this is what I have firmly decided: 9 If you do not tell me the dream, there is only one penalty for you. You have conspired to tell me misleading and wicked things, hoping the situation will change. So then, tell me the dream, and I will know that you can interpret it for me.”

    10 The astrologers answered the king, “There is no one on earth who can do what the king asks! No king, however great and mighty, has ever asked such a thing of any magician or enchanter or astrologer. 11 What the king asks is too difficult. No one can reveal it to the king except the gods, and they do not live among humans.”

    12 This made the king so angry and furious that he ordered the execution of all the wise men of Babylon. 13 So the decree was issued to put the wise men to death, and men were sent to look for Daniel and his friends to put them to death.

    14 When Arioch, the commander of the king’s guard, had gone out to put to death the wise men of Babylon, Daniel spoke to him with wisdom and tact. 15 He asked the king’s officer, “Why did the king issue such a harsh decree?” Arioch then explained the matter to Daniel. 16 At this, Daniel went in to the king and asked for time, so that he might interpret the dream for him.

    17 Then Daniel returned to his house and explained the matter to his friends Hananiah, Mishael and Azariah. 18 He urged them to plead for mercy from the God of heaven concerning this mystery, so that he and his friends might not be executed with the rest of the wise men of Babylon. 19 During the night the mystery was revealed to Daniel in a vision….


    Comment by erdman31 — 30 June 2011 @ 11:05 am

  3. It’s further interesting to me to compare the phenomenon of dream interpretation. Both in ancient times and in these post-Freudian modern days, the power to interpret dreams is a position that comes with a good deal of respect and financial benefit. We feel a good deal better if someone can organize the chaos that troubles us in the night.


    Comment by erdman31 — 30 June 2011 @ 11:08 am

  4. I agree that dreams and their interpretation are intriguing, Erdman. However, I must point out that the pond and my mistaken perceptions happened while I was awake — a real pond, with real scum on most of its surface, except for a real patch of clear water. Now I suppose you could argue that my interpreting the trompe l’oeil effect first as cow, then as horse might reveal something of my unconscious drives and desires and fears. I already knew that the field surrounding the pond is frequently grazed by cattle and occasionally by horses, so in context those seemed like the first best guess.

    Would others happening to pass by this same pond have realized immediately that what I saw as a large object in the pond was in fact an irregular patch of clear surface water surrounded by pond scum? If I’d been running with someone else I could have asked him what s/he saw, which might have helped me clarify my confusion more quickly. Did I have an unconscious desire to see something there other than just a static small body of water? Seeing a head on a long neck emerging from water: was this a primal phallic image rising up in my unconscious that I projected onto the scene? I suppose it’s possible. In the same spirit, what was there in my post that caused you too see it as the description of a dream rather than of a scene in the world I saw with my open eyes during broad daylight? What was there in your unconscious that replaced a description of a real scene with that of a dream?


    Comment by ktismatics — 30 June 2011 @ 11:22 am

  5. Here’s a more textbook example of the role of inference in perception. You see a dog running along behind a white picket fence. All you ever see are regularly-spaced vertical stripes of the dog interspersed by white stripes of fence, yet you have no trouble assembling a whole continuous moving object out of those stripes. Or I hear two bird songs, identical in pitch and melody, one louder than the other: I have no trouble recognizing that there are two separate birds, one closer than the other, making those sounds.


    Comment by ktismatics — 30 June 2011 @ 11:30 am

  6. Misperceptions are sort of like dreams — the distortion, the image that doesn’t quite jibe with what one already knows about reality. I suppose that misperceptions, optical illusions and so on contributed to the theory that the real is constructed, that reality is inside the head rather than what the head is inside of. But the implicit normativity of perception, iteratively matching input with inference, must be acknowledged. Just because I perceived a horse didn’t make it a horse; it really wasn’t a horse.


    Comment by ktismatics — 30 June 2011 @ 12:03 pm

    • It’s pieces like this that made me mistake you for an Eng Lit academic. Strong and evocative.


      Comment by W.Kasper — 30 June 2011 @ 7:23 pm

  7. Thanks. A psychologist though. Today I was wondering whether it’s my schizoid tendencies, my loose hold on realities both social and physical, that make me more susceptible to optical illusions. Also to interpersonal illusions.


    Comment by ktismatics — 30 June 2011 @ 7:25 pm

    • Falling in love’s a big interpersonal illusion, so it’s not all bad. I find the more ‘alert’ I am to social realities, the more distant I become on an interpersonal level. A ying/yang thing maybe?


      Comment by W.Kasper — 1 July 2011 @ 12:39 am

    • Is there a distinction between real love and illusory love? I’d say yes. Must real love be discovered, or is it really possible to make love? I’d say yes to that too, but it takes at least two to tango. Though there are surely panpsychists and object-oriented metaphysicians who would insist that love between two billiard balls is not only possible but actual, this is almost surely bullshit.

      Yesterday I was discussing with our daughter exactly what you’re saying about the simultaneous alertness to and distance from social realities. We tentatively agreed that this alertness entails a fascination with the alien that can be achieved especially well by the already-alienated. This alertness is useful in art and writing, as well as in empirical and analytical investigations of the interpersonal. We find, however, that we’re more interested in satisfying our curiosity about others than in bonding with or helping them. This morning I asked my wife about this distanced interpersonal POV: she thinks that my distance enables me to entertain more possibilities about others’ perspectives before foreclosing them in honor of common knowledge or intuition.


      Comment by ktismatics — 1 July 2011 @ 12:09 pm

      • It’s interesting – to me anyway – how ‘wider world’ realities feel more distant when I’m in more ‘involved’ social circles. When that’s less of a concern, the wider world suddenly feels more upfront. Felt like a lot of historical/political events piled up that I’ve only been able to really take in getting more time to myself relatively recently. It’s like perceiving a friend’s eccentricities more sharply when you haven’t seen them for a while.

        But yeah, real love is more to do with unconditional attachment and commitment. Illusory love can be like getting wrapped up in TV show box sets. When you’ve finished watching them all week, it’s like “hey – what happened? That show wasn’t that good anyway.” Final episode syndrome.

        Whatever billiard balls may feel about each other is surely a private matter. There’s no way of eavesdropping on their pillow talk anyway.


        Comment by W.Kasper — 2 July 2011 @ 4:47 am

      • Do you believe that romantic love and family and friendships are social constructs created/manipulated by the politically and economically powerful, narrowing people’s libidinal investments so that they don’t spill over into larger social-libidinal affiliations which might pose more of a threat to the status quo? Certainly the bureaucratization of government enhances, probably intentionally, the alienation of the governed, while media coverage distances politicians further by enhancing their celebrity status. Maybe in GB it’s possible to find some circuitry of political collegiality that enhances rather than depletes libidinal energy, that reinforces a sense of mutual empowerment rather than futility.


        Comment by ktismatics — 2 July 2011 @ 9:09 am

      • Romantic love, family and friendship are social constructs – but that doesn’t necessarily prevent them from being authentic, meaningful emotional bonds. The problem may be when those bonds are psychotically focussed ‘inwards’ and become a kind of default weapon against others, denying the legitimacy of ‘outsiders’ (from religious bigotry to things like families ‘closing ranks’ on in-laws etc). I’m close with relatives, but I’ve known families that are inner-directed to a very unhealthy degree (ie. it’s very uncomfortable and antagonistic when they gather together – an almost ‘military’ kind of groupthink). I think stories of royal intrigue, dynasties and mafia dramas play on that tension.

        Watching the way governments, technology, education and corporations have changed social interaction in recent years (with war and finance agendas always lurking behind it), I’d say they probably do have an investment in severely narrowing the meaning and ‘authenticity’ of social bonds. Seems to be enhancing alienation and moving wider ‘libidinal investment’ in a (paradoxically intense) ‘teflon’ direction – certainly in the UK, which is small geographically, very media-driven and subject to more concentrated power blocs. It doesn’t feel ’empowering’ – quite the opposite, I’d say – although the explosion in comms tech and 24 hour reportage tries to suggest otherwise. The narrowing focus on ‘families’ in mainstream rhetoric (and by extension ‘the community’) can be very dangerous, divisive and ugly if it’s ‘pure’ example (nuclear white hetero family, in full working health, on median income and home-owning) is treated as the only one worth respecting or communicating to.


        Comment by W.Kasper — 2 July 2011 @ 12:04 pm

      • The narrowing focus on ‘families’ in mainstream rhetoric (and by extension ‘the community’) can be very dangerous, divisive and ugly if it’s ‘pure’ example (nuclear white hetero family, in full working health, on median income and home-owning) is treated as the only one worth respecting or communicating to.

        That’s all o.k. but isn’t Facebook now a revolutionary power and isn’t “social” one of the central buzzwords of our time, orders of magnitude ahead of “familiy values” or “grizzly mums” ( I guess there is no such thing in Europe simply because there are no grizzlies )?

        What I’m asking for is the relevance of your attack vector against the very much promoted and even more hated white, bourgeois, hetero core family, the origin of neurosis and even worse things like individualism and capitalism. I sometimes wished that China becomes the superpower everyone is promising now and with it the rise of a rude capitalist collectivism and biopower – if for nothing else but a refreshed social analysis in the self-centered West which seems to be stuck with the schemes it worked out in the late 19th and early 20th century.


        Comment by Kay — 3 July 2011 @ 3:32 am

  8. The causal features of the objects themselves act as constraints upon the development of our dispositions to respond to them in perception and action, and via them upon our dispositions to reason about them.

    Sure, but don’t we want to know what those “dispositions to reason about” exactly are s.t. we can remove ourselves from them? In the “cognitive science turn” we think about thinking objectively/algorithmically s.t. it could also be done by a machine. There shall be no need to invoke god or subjectivity in the stream of processes which leads to object recognition, classification, re-classification and class-building. Nothing of it relies on consciousness.

    But what is consciousness anyway? The problem we face in the objective description of the mind is that at some point we will talk about implementation details like brains, massive parallel processing of trillions of neurons which are active and shuffle information around etc. At the same time there is just one body, brain, vision and self. So there is a vast network of decoupled processes which occasionally get synchronized and at the same time they are all one – a single stream of thought and perception which serve the purpose of maintaining the oneness of the body-self in the world. In mysticism this oneness can grow beyond the body to finally include the whole universe. Is this wrong? Since we don’t understand how a myriad of distinct processes can be one, not only as a formal closure in the sense of set building, but in a real, experiential sense, on which base could we judge? All this “philosophy of mind” and cognitive science with their thousands and thousands of approaches and treaties and scholasticism and not a single question can be answered and assertion refuted.


    Comment by Kay — 1 July 2011 @ 12:04 am

    • “treaties”


      “There shall [sic] be no need to invoke god or subjectivity in the stream of processes which leads to object recognition, classification, re-classification and class-building. Nothing of it relies on consciousness.”

      So a machine can recognize loathsome, wizened shit like you write? I’d as lief read Levi Bryant at this point. He was always at least somewhat superior to KVond as well, and knows how to write proper English. There’s ‘I shall,’ ‘We shall’, and sometimes the special ‘This, too, shall pass’. Probably the ‘There shall be no need…’ is written not only because of your usual bilingually-challenged writing, but as a kind of Prussian-style order, gaming as always and probably not even in Munich. And who would care? Your disposal of the idea of place is effectively null and void, but it was a good try.

      But I agree, no ‘god or subjectivity’ is needed to adjudge your tripe. A machine may not be able to recognize subtleties, but s/he can certainly recognize the overtly repulsive–somewhat like tech support diagnosing hardware problems s/he can’t see even through remoteservices.


      Comment by Illegal dances of New York City — 1 July 2011 @ 11:06 am

  9. “In the “cognitive science turn” we think about thinking objectively/algorithmically”

    Pete illustrates the move beyond subjective perception with the case of the colorblind person. This person could rely either on other people’s perceptions or on machine readouts of light wavelengths in order to gain an objective understanding of color. Scientific investigation in general attempts to gain understanding of the world that isn’t contingent on the limitations and biases of so-called folk psychology. And as you say scientific objectivity can be applied to human cognition as well. More traditional empirical psychology investigates consciousness, belief, and so on by assuming that these constructs are real. Eliminative approaches pursue understanding from the bottom up so to speak, regarding the distributed network as the “real” of cognition while the subjective experience is merely a folk-psychological illusion. I remain agnostic and ambivalent about the relative merits of these two approaches. I would say, though, that while “cognitive psychology” might from the outsider’s perspective be engaged in understanding “the mind,” actual practitioners in the field focus their empirical and theoretical work more narrowly on, e.g., problem-solving and recognition memory. Big constructs like “the mind” tends more to be invoked by philosophers. Empirical work conducted at a more detailed level tends not to interest the philosophers, while the empiricists disregard the big “philosophy of mind” questions as beyond verification or falsification by evidence. The same I think would be the case with concepts like “the real” or “objects” or “processes”: these are too abstract for empirical scientific work, whereas specific scientific findings tend to bore the metaphysicians.


    Comment by ktismatics — 1 July 2011 @ 11:49 am

    • Eliminative approaches pursue understanding from the bottom up so to speak, regarding the distributed network as the “real” of cognition while the subjective experience is merely a folk-psychological illusion.

      That’s o.k. as long as we agree that the “mind” shall not explain vision or language understanding. Things get more interesting though when perception, language, and action are coordinated s.t. that there is a single coherent stream of events ( our own tale ) having an internal, sometimes screwed logic ( meaning / sense ) which can be abstracted from the lower layers which perform the basic processing. Having no better term I’d call this a “mind” and I don’t believe that neurons and biochemistry are a necessary condition for it to exist.

      Empirical work conducted at a more detailed level tends not to interest the philosophers, while the empiricists disregard the big “philosophy of mind” questions as beyond verification or falsification by evidence. The same I think would be the case with concepts like “the real” or “objects” or “processes”: these are too abstract for empirical scientific work, whereas specific scientific findings tend to bore the metaphysicians.

      This sounds like both parties peacefully go there ways and hardly ever interfere. My impression is very different even though they do not work in the same department.


      Comment by Kay — 2 July 2011 @ 1:22 am

      • ‘screwed ‘


        Interesting that you find obvious self-sabotage as witty as ‘your own tail’.


        Comment by Illegal dances of New York City — 2 July 2011 @ 8:11 am

  10. “the “mind” shall not explain vision or language understanding”

    Do you refer to the naive-psychological notion, a variant of mind-body dualism, that a homunculus called “mind” assembles inputs from remote sensory streams in order to assemble meaning of visual and linguistic inputs? If so I agree. If we accept the idea of “embodied mind,” such that at least the human mind consists of the coordinated action of the senses, the neural pathways, the nodes and synapses in the brain and so on, then perhaps mind eventually can achieve explanatory power. But i also agree that vision need not rely on the kind of embodiment that humans or other animals rely on. Machines can do it using very different hardware.

    “This sounds like both parties peacefully go there ways and hardly ever interfere.”

    If truth is more than a social construct, then one should expect that the truths of philosophers and those of scientists from varying disciplines would converge when addressing the same sorts of questions. That was one motivation for writing this post: empirical observations in visual perception support Brandom’s philosophical idea that iteration between observation and inference progressively reduces error and achieves a more accurate representation of the world.

    One could imagine an alternative interpretation of the visual experience I describe in this post: Two new objects emerged from the interaction of my visual perceptual apparatus and the pond. The first object was a perceived giant horse; the second, a perceived reflection of a partly cloudy sky. Neither emergent object is a more true representation than the other; both can be evaluated for their utility, beauty, affective impact, and so on. I can say with some assurance that empirical researchers of visual perception would not find this alternative interpretation of optical illusions very useful in their work.


    Comment by ktismatics — 2 July 2011 @ 10:41 am

  11. “The psychic activities, by which we arrive at the judgment that a certain object of a certain character exists before us at a certain place, are generally not conscious activities but unconscious ones. In their results they are equivalent to an inference, in so far as we achieve, by way of the observed effect upon our senses, the idea of the cause of this effect, even though in fact it is invariably only the nervous excitations, the effects, that we can perceive directly, and never the external objects. Nevertheless, they thus appear to be differentiated from an inference, in the ordinary sense of the word, in that an inference is an act of conscious thinking. There are, for example, actual conscious inferences of this sort when an astronomer computes the positions of the stars in space, their distances from the earth, etc., from the perspective images he has had of them at different times and at different points in the earth’s orbit. The astronomer bases his conclusions upon a conscious knowledge of the laws of optics. In the ordinary acts of seeing, such a knowledge of optics is lacking; still it may be permissible to designate the psychic acts of ordinary perception as unconscious inferences.”

    – Hermann von Helmholtz, Optik, Volume 3, 1866

    In his 1950 History of Experimental Psychology, Edwin Boring emphasizes that, for Helmholtz, “Unconscious inferences are formed by experience. Thus the doctrine of unconscious inference becomes the tool of empiricism.” The processes by which experience shapes inference and vice versa, as well as the role of instinct in the process, has been the subject of ongoing investigation. However, the understanding of perception as an unconscious iterative convergence on accurate representation has persisted in perceptual psychology since Helmholtz’s original formulation.


    Comment by ktismatics — 2 July 2011 @ 12:09 pm

    • I expect that an adult human brain performs fast categorizations in one pass on average. Multiple passes or iterative convergence ( what happens when convergence fails? ) may more likely be conscious: for example one varies ones own position for a better view or asks oneself if this is really the object one guesses to see and gathers more information to make a yes/no decision. This may not come as a surprise because multiple passes require attention during an extended time span. Therefore the mind is invoked to observe, control, adjust or reject the whole process based on its meaning.

      To clarify what I said above: the mind doesn’t contribute to fast analysis of images or language. I’m not aware of languages syntactic or semantic analysis and whether it is done using built in grammars ( Chomsky ) or statistical inference ( Norvig ) or some hybrid approach ( would be interested in working this out ). But if this is going to fail, consumes much time and requires attention ( why? ), the mind is going to contribute to the problem resolution, possibly by extending itself using tools. If attention can be withdrawn it might happen that the brain can continue working on a problem solution and writes it to memory. Next time we think about it, we are surprised the solution is just present.


      Comment by Kay — 3 July 2011 @ 12:11 am

      • Whoever you are ‘Kay’, we’re not computers and we’re not facebook pages. They didn’t figure much in anyone’s life when I was in nappies, or even combat trousers. I doubt they mean much to majority of Chinese either, despite your crypto-racist cyber-fantasies.

        The whole ‘brain as software’ doo-doo is a dated 90s fraud, as much as ‘financial innovations’ bringing everlasting prosperity. Cyborgs and derivatives are sooo generation X. But I guess some people refuse to let go of their youth. Put your Jeff Mills mixtapes back in the attic.


        Comment by W.Kasper — 3 July 2011 @ 3:19 pm

      • I’m not generally subject to the allure of the post-human borg singularity. I’ve inferred in some circles the valorization of a post-humanist renunciation of love and family in order to clear the way for Reichian swarming energy and collective political discipline, but I try to be careful not to create a caricature of such positions. It’s clear though that vision isn’t limited to humans. Animals can see, and so can machines. Machine vision operations are based on that of humans and other animals, via iterative observe-infer-feedback cycles by which the machine’s neural net builds up an increasingly more accurate representation of the environment. Robots with machine vision can find objects while avoiding obstacles along the way — I recently watched a bomb-detecting robot being tested at a local school that demonstrated these abilities. Machine vision works better now than it did in the 90s, though clearly whatever automation of seeing they’ve achieved hasn’t accrued to shorter work weeks or higher pay for the replaced human seers. And the deployment of machine vision devices in the air above places like Gaza certainly has added to the paranoiac sense of perpetual surveillance characterizing the 21st century.


        Comment by ktismatics — 3 July 2011 @ 3:53 pm

      • @ W. Kasper. An anecdote from the ’90s. At that time many people in AI research showed some interest in biology inspired computing models. Those were generally good at solving a few problems like clustering data ( object recognition ) or solving global optimization problems but lacked universality. For example one couldn’t even do simple arithmetics with them or model stacks and procedure calls, which are required to parse language in the way Chomsky imagined in the 1950s. I went to my university library for introductory literature in biology: evolution, cell and molecular biology and of course also neurosciences. It was great fun. To my surprise the biologists already used the computation models inspired by their own work. For example there was a neuroscientist who made extensive use of self-organized Kohonen networks to explain feature mappings. We’ve created a full circle. Everyone admitted that Searle was right, that humans are not computers, that the software metaphor of the mind was flawed that syntax is not semantics. Mysterious “causal powers” in the brain were suggested instead but those who were dedicated to establish those “causal powers” happily used our computation models, which reflected their work.

        I’m also aware of the cyborg-mania in social theories of that time and I admit that I liked Hans Moravec’s prophetic speculations based on Moore’s law. The singularity was pop, like Stephen Hawking, techno music, chaos theory and the love parade. These days we have social media. Everyone is turned into a Facebook page, the personal features of everyone including philosophy bloggers are dissected in painful detail and we got a full bestiary of trolls, vampires and crackpots – a complete victory of the humanists over the techies who were just told to not feed them and not care much about people features otherwise. While dissent about ideas and opinions becomes immediately personal, individualism declines, which I find a bit sad. So much about this.


        Comment by Kay — 3 July 2011 @ 11:19 pm

  12. No doubt it’s true that one pass only is required for most adult perception by humans, and by other kinds of animals for that matter. Because perception is so automatic and so nearly instantaneous, it seems as though perception is either direct or innate. That’s why studies of infant perception are important: are what Helmholtz termed “unconscious inferences” actually direct perceptions of the world, or are the inferential mechanisms innate to the brain, or are they learned? These are the sorts of questions that drive empirical investigation. Declaring oneself an epistemological nativist or a positivist or an associationist without looking at what perceptual scientists have been discovering for the last 150 years — is this sort of thinking not unlike drawing inferences about the world without actually looking at the world?

    For example, adults “infer” a 3D scenario from the 2D pattern of sensory inputs derived from the world. Is this inferential ability present at birth, or there an interval during infancy when the human fails to make this transformation? So-called visual cliff experiments are conducted, in which an infant is placed on a flat surface covered in a checkerboard pattern. The surface is continuously flat and solid, able to sustain the infant’s weight, but there is a well-defined edge across which the checked surface drops down three feet or so. The drop-off is visible from above through a transparent surface, creating the optical illusion of depth. Here’s a photo of the apparatus. It turns out that, before kids learn to crawl, they show no distress when pushed across the surface and over the visual cliff. Once they achieve self-locomotion at around 6 months of age they avoid going over the cliff and show distress when moved across the line. This is one source of empirical evidence supporting the contention that 3D perception is learned rather than innate, even though it occurs automatically most of the time in adults. Other kinds of creatures show the same learning: e.g., rats begin to self-locomote at 4 weeks, which is when they begin to avoid the visual cliff. The nativist would contend that it’s not empirical experience but rather maturation of the innate capability that determines when the creature begins demonstrating 3D visual perception. This too becomes an empirical question.

    Language acquisition is a more clear-cut example of how iterative incremental learning is compiled by the human child into one-pass, unconscious inferences of the meaning of linguistic utterances. Clearly humans aren’t born with the ability to use language, but by the time they’re 3 years old they’ve become quite adept. It seems that this very distinctively human ability is achieved unconsciously, or at least without conscious self-reflection, during early childhood interaction with other language users. There is empirical evidence showing that adults unconsciously pick out and understand certain bits of language mingled in with background noise, suggesting that linguistic inference from auditory sensation eventually becomes automatic. Here again we see the debates between those who regard learning as critical versus those who argue for maturation of the “language instinct.” I’ve posted previously some research supporting the learning approach.

    Toward the end of Part Four of his post, Pete begins to critique the idea that human percepts determine the content and meaning of human concepts:

    “The idea here is essentially the representationalist one that our practices for perceptually classifying and practically coping with what the word refers to determine what the concept represents, and that this determines its inferential role, rather than the other way around. However, this doesn’t work if we consider the issue in more detail.”

    This remark, juxtaposed with my specific experience of a visual illusion of a giant horse in the pond, prompted me to write this post. Even before a human perceiver assigns the word “jellyfish” to that object floating on the sea, that human must distinguish the jellyfish as a 3D object distinct from the rest of the 2D surface of the sea. Inferential processes are already at work, unconsciously and non-linguistically, in transforming the 2D visual sensory array into a 3D scenario consisting of surfaces, depths, and discrete objects. I.e., perception is not direct, but perceptual representation isn’t linguistic either. The experiential-inferential iterations go “all the way down,” even to non-humans.


    Comment by ktismatics — 3 July 2011 @ 7:39 am

  13. It’s worth noting that visual perception need not rely on inference prompted by memory of specific scenes or categories of objects. JJ Gibson and other advocates of “ecological perception” contend that the information provided by a well-lit environment is fully adequate to form an accurate visual representation of that environment, whether that representation be assembled by the eyes and neural pathways and brains of animals or by machine sensors and algorithms. The perceiver, it is argued, need only learn to fine-tune the perceptual apparatus, much as the user of a telescope must learn how to focus the device in order to get a clear view. Even optical illusions can be resolved through better lighting and movement relative to the viewed object. It’s clear that perception is always representational: light is captured by rods and cones and passes through the optic nerve to the brain. However, according to the Gibsonian interpretation the visual environment is always the gold standard for evaluating the accuracy of the perceptual representation; mental inference is relegated to an even less central role than that accorded by Helmholtz and other inferential theorists of perception.


    Comment by ktismatics — 6 July 2011 @ 4:31 pm

RSS feed for comments on this post. TrackBack URI

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

Blog at WordPress.com.

%d bloggers like this: