Subscribe to:
Post Comments (Atom)
-
Harnad, S. (2003b) Categorical Perception . Encyclopedia of Cognitive Science . Nature Publishing Group. Macmillan. http://www.youtube.com...
-
(10a. Comment Overflow) (50+)
-
Psychology PSYC 538, Winter 2018: Categorization, Communication and Consciousness Time : Tuesdays 2:35-5:25 Place : STBIO S3/3 ...
Watanabe’s Ugly Duckling Theorem, where other ducklings look more similar to each other than the swanlette is explained by the fact that some features become more salient than others, as they are differentially weighted. This selectivity that arises as a consequence of the visual system is described in the text as being arbitrary. I find this use of the word, arbitrary as relatively ambiguous since it could be interpreted as applying to the visual system’s selection of these features or as a consequence of other influences such as hearsay or learning. By this I mean, rather than these features arising from arbitrary abstractions performed by our visual system, perhaps the features more heavily weighted in this case are those made explicit by others in the community who make these features explicit by expressing them as invariant. Nonetheless, describing this phenomena as an example of hearsay may be circular as hearsay fails to explain the process of category learning all the way down and requires sensori-motor interactions, hence, the arbitrary abstraction by the visual system. This circularity is avoided if we account for the phenomena as being a type of Whorfian effect, thus being shaped by language and culture.
ReplyDeleteIn class, we often simplify categorizing as "doing the right thing with the right type of thing". This reading was very interesting to me as it read almost like a list of premises building to the overall conclusion that cognition is categorization. Specifically, I'd like to delve more into the proving of each premise, as supported (or contrasted) by various studies I've learnt. Categorization is first defined in this paper as "any systematic differential interaction between an autonomous, adaptive sensorimotor system and its world". Though this definition is much more complicated than our in class one, I do think both say approximately the same thing. That is, "any systematic differential interaction" would be equivalent to "doing the right thing" and "between an autonomous, adaptive sensorimotor system and its world" identifies what qualifies as the right types of things. I found the reasoning behind each of "systematic" and "differential" to be very interesting as well. My only issue would be that it is stated that what does and doesn't exist should be left to specialists in the various fields. While I do agree that for most concrete things (food, animals), I think it is much harder when the topic for the category gets more abstract. For example, there are still those who don’t believe in homosexuality as a concept within the sexuality category. I think that to some extent, a lot of categorization is highly subjective, and specific to the opinion of the person.
ReplyDeleteIn regards to the innate ability to respond differentially to things, the paper seems to side against nativism. There was a study by Landau, Smith and Jones which showed that there were innate biases that young children display when categorizing things. There was a bias towards shapes, but not towards size or texture. If the shape of an object is changed, the child will no longer apply the same word for it, implying that they think shape determines the members of a category more than size and texture. While this isn't completely against nativism (one could argue that they had just learnt to do this very early on in life, before the study) it does seem to support that we have some innate biases when categorizing. Lastly, from a linguistic side, categorical speech perception also seems to be innate. You can create a spectrum, slowly increasing voice onset time, which is the key differentiating piece between the phonemes [ba] and [pa]. The perception of listeners will not be a straight line as one might expect. As voice onset time increases, there is a very strict boundary as to where people stop hearing [ba] and start hearing [pa] rather suddenly.
“They were concerned only with describing the input and the sensorimotor interactions, not how a sensorimotor system could actually do those things. So whereas they were already beginning to scratch the surface of the "what" of our categorization capacity, in input/output terms, neither was interested in the "how." “ (page 4)
ReplyDeleteThis reminds me of Fodor’s article, where his main fault with the neuroimaging studies is that they mainly discuss what is occurring and not how those brain processes occur, since it is all correlational research.
“Although basic color CP is inborn rather than a result of learning, it still meets our definition of categorization because the real-time trial-and-error process that "shaped" CP through error-corrective feedback from adaptive consequences was Darwinian evolution.“ (page 6)
ReplyDeleteThis reminds me a bit of the Whorf hypothesis, which suggested that speakers of languages without a name for each colour would not be as adept at categorizing and identifying distinct colours. In the case of the passage quoted above, it suggests that we need feedback in order to correctly categorize colours. I can agree that there may be some need for feedback in learning the name of colours, but I can’t see why an individual would need corrective feedback to identify colours, unless of course they were colour blind. In the same way that I would never mistake a navy for baby blue, there are many other shades of blue that I do not know the specific name of, but that I can nonetheless identify as unique and distinct.
This comment has been removed by the author.
ReplyDeleteI found the concept of vanishing intersections a little confusing. What I understand is that it is supposed to be applied to categories that don’t seem to have any common invariances, and because of this Fodor argues that language acquisition was to be innate , because kids are able to learn categories with no shared invariances. However, I neither agree with this thought nor do I understand the purpose of the example of the "shadows of boomerangs or tables or chicken bottoms."
ReplyDeleteI appreciated the definition of categorization as any systematic differential interaction between an autonomous, adaptive, sensorimotor system and its world; the paper then goes on to explain that essentially categorization can be thought of as non-arbitrary interactions, that are based on the same KIND of input producing the same output, and a different kind of input produces different outputs, and that it must have the quality of not producing the exact same output across time, every time. I would say that, a thing I wished we dived more deeply in this course was a focus on defining and fully understanding the concepts that make up the definition of a larger concept (such as one must first know what adaptive, systemic and differential means in the context of this paper before they truly understand categorization)
Likewise the definition of learning as a system that samples inputs and generates outputs in response to them on the basis of trial and error, guided by corrective feedback was a really nice way of explaining the concept.
I believe you once talked about how categories are discrete but at times they can be continuous. What I understand this to mean, is that when you are dealing with the color black and white they are two separate categories, yet if you have gradients of grey, where some of them are 70% black, 30% white, others 70% white, 30% black and so on, you can still end up categorizing the 70% black as black and the 70% white as white, although the colors are in fact grey (which is an intermediate of the two) , so in this way the spectrum is continuous (all the gradients of grey) but essentially it is discrete (either black or white)
I was surprised that language was only mentioned towards the end of the reading, considering how crucial it is to categorization and our ability to learn to categorize quickly and efficiently. It is also interesting that both language, which facilitates categorization, and categorical perception (CP) are innate, Darwinian processes, the lack of which led to dying out. This tells me that both language (and in turn, categorization) and CP are extremely important processes - cognition is categorization - and that if there were a robot that could pass T3, it would have to have these three abilities (language, categorization, and CP). I am aware that computers are modelled with "learning" processes that exemplify feedback and trial and error. Could it be possible that what is missing from computer modelling is the threat of extinction, or the powerful force of natural selection? Regardless of whether an individual is aware of it, natural selection does influence one's fitness. I wonder if robots were to experience this pressure, they would be more indistinguishable in all facets. Not only would they learn and adapt to their environment, but the ones that did not would die out. Also, I found it particularly interesting that our sensory system chooses which features are more salient and weights them differently without our conscious awareness. Perhaps this contributes to our inability to explain what we are doing when we are categorizing. If we cannot consciously describe what features we are detecting and choosing to sort and why, how can we even begin to explain how we are categorizing?
ReplyDeleteThis kind of extreme nativism about categories is usually not far away from something even more extreme than nativism, which is the view that our categories were not even "learned" through evolutionary adaptation: The capacity to categorize comes somehow prestructured in our brains in the same way that the structure of the carbon atom came prestructured from the Big Bang, without needing anything like "learning" to shape it.”
ReplyDeleteSince I’m a big fan of Chomsky and his nativist approach saying that we are born with the innate ability to develop language I found these particular phrases interesting. I always find the battle between nativism and tabula rasa interesting and categorization could be argued as being in one or the other. Although with our evolution and need to survive as humans it would make sense that categorization is innate and pre-structured in order to categorize and understand things in our life, it would make sense that categorization is learned. When children see a Chihuahua they will say doggy instead of saying Chihuahua but even more might recognize a dog as an animal before they can differentiate a dog and a cat. By learning and trial and error categorizing is improved. Furthermore the categorical perception element of categorization and cognition makes sense with the same basic example I provided about the 3 different levels of categorization (basic, superordinate and subordinate) since a child could see blood orange and see it as a color first, then learn that it is red and then learn about saturation, brightness and hue and possibly see that it is blood orange. These all come from learning and not innate abilities.