Harnad, S. (2003b) Categorical Perception. Encyclopedia of Cognitive Science. Nature Publishing Group. Macmillan.
Differences can be perceived as gradual and quantitative, as with different shades of gray, or they can be perceived as more abrupt and qualitative, as with different colors. The first is called continuous perception and the second categorical perception. Categorical perception (CP) can be inborn or can be induced by learning. Formerly thought to be peculiar to speech and color perception, CP turns out to be far more general, and may be related to how the neural networks in our brains detect the features that allow us to sort the things in the world into their proper categories, "warping" perceived similarities and differences so as to compress some things into the same category and separate others into different categories.
Pullum, Geoffrey K. (1991). The Great Eskimo Vocabulary Hoax and other Irreverent Essays on the Study of Language. University of Chicago Press.
At the end of Harnad's 'Categorical Perception', the paper veers back towards the grounding problem. It argues whether CP can ever be based in language rather than sensory-motor experiences.
ReplyDeleteBased on our discussions in class it seems clear that the initial phases of categorical perception rely entirely on sensory-motor experiences. The question remains, however, at what point can we take the language that is already grounded in sensory-motor experiences to then create new categories with just language. At that point can it be argued that the CP is then free of sensory-motor input? I would argue that it is not, since it still must rely on language already grounded in that input. Even after we create categories several degrees removed from the sensory-motor experiences that first created the baseline categories, the newly created categories are in a way still dependent on those experiences.
I went to a talk prof. Harnad gave a long time ago and I seem to recall him saying that with language you get to a combinatorial explosion once you gain a critical number of (sensory-motor) categories. From there you can build up into more and more "ungrounded" and abstract categories. I may be wrong though. He'll likely go through this in the future lectures.
DeleteMy understanding is that the higher “ungrounded” & abstract categories are actually grounded. Think of them like high up branches of a tree. The highest branches of the tree do not touch the ground themselves, but they are connected through a system to the roots that hold them up high in the sky, with the roots being the “lower” sensory-motor grounded categories. If there are no grounded roots, there could never be higher branches.
DeleteFor a real world example, think of the abstract idea of justice. There is no single referent that one could point to that could holistically define justice, but there are millions of small events that could be grounded (i.e. watching someone break into a storefront and stealing money from the cash register) that all form their own category that is justice. Categories such as money or a store might need to be grounded, which then interact in the case of a robbery that leads to an understanding of justice. If one has no knowledge of smaller, ungrounded events like the one mentioned above or if they lack an understanding of basic concepts in the event, there is no way that they could understand justice. To be clear though, one could learn about these lower categories via instruction, not just by witnessing them (induction).
On this train of thought, I have trouble with the special status of language. If cognition is simply categorisation, then language does not seem necessary to cognition. But the whorfian hypothesis suggests that human capability necessitates language. Are we saying that language necessarily accompanies categorisation? So that no categorisation can happen without language co-occurring? What kind of computational structures must underlie such a distinct linguistic capability: to put names to categories?
DeleteRe: Can categories, and their accompanying CP, be acquired through language alone?
DeleteIs Harnad suggesting that it conclusive to state that once a minimal set of categories are “grounded” through sensorimotor experiences (and some through evolution) then we have access to acquiring all categories (both in the real world and not) through language alone?
I interpreted this in the same way as Annabel. Maybe evolved CP and learned CP, although both necessary are insufficient in explaining categorization.
Delete"Both innate and learned CP are sensorimotor effects: The compression/separation biases are sensorimotor biases, and presumably had sensorimotor origins whether during the sensorimotor life-history of the organism, in the case of learned CP, or the sensorimotor life-history of the species, in the case of innate CP."
It seems that both evolved and learned CP are somehow grounded in sensorimotor experience however, as pointed out in the article, concepts like truth and unicorns present a challenge to this notion. I think that while innate CP, evolved CP are the basis of categorization, language abilities are necessary for acquiring "higher" concepts like unicorns and truth. Monkeys for example have both innate and evolved CP but without linguistic abilities, would they be able to understand the notion of truth?
It is also worth noting that sensorimotor capacities only allows us to discriminate between categories, while language allows us to identify and manipulate them. This is why higher level and abstract categories, which do not need to be perceptually discriminated, would be acquired through language alone: they are created (and thus identified) via our linguistic abilities.
DeleteAdrian Many of you are mixing up categorization, category learning and categorical perception.
DeleteCategorization is doing the right thing with the right kind of thing. Category learning is learning to do the right thing with the right kind of thing. And CP happens when there is compression of similarities within the same category and separation of differences between different categories: CP can be innate or a result of learning, but it is a kind of enhancement of category differences and distinctiveness, so it is less likely to happen when the category is easy and the differences are obvious.
Your question is not about CP but about sensorimotor induction versus symbolic (verbal) instruction: Instruction has to be grounded by induction. Words are just recombinations of grounded category names in propositions defining new (combinatory) categories. About 1500 categories grounded by induction seems to be enough to define all the rest verbally.
Auguste No combinatorial explosion once you gain a critical number of (sensory-motor) categories (about 1500). Just the capacity to express all the rest by combining the ones you already have into propositions.
Karl New categories can be directly grounded by induction even when you could have done it verbally.
Yi Yang New categories can be learned by verbal recombinations of already grounded categories.
Annabel In principle you could ground a 1500-word minimal grounding set (T3), go into a room, and learn all the rest verbally (T2). I'm not saying it's practical possible to do that, or that anyone does. I think we do direct induction throughout our lives, but words cut down enormously on the need for that -- and they also open up recombinatory categories that can only be learned from words. Of course for any brand new sensory experience -- a new taste very different from anything we've tasted before for which there is no word -- induction is the only way.
Elise Concepts like truth and unicorns don't present a challenge, if you have the grounded words to define or describe them.
Monkeys could learn if they can trust you, but the notion of truth is a propositional one, and they have no propositions.
Mael :
Discriminate = perceive whether two (simultaneous) things are same or different.
Identify = to name (categorize).
Propositions allow you to recombine grounded categories. Every sentence is about categories: The subject category is inside the predicate category. (Apples are red things.)
RE: “but to show that it is a full-blown language effect, and not merely a vocabulary effect, it will have to be shown that our perception of the world can also be warped, not just by how things are named but by what we are told about them”
ReplyDeleteA universalist would argue that 1. evidence suggests that categorical perception (CP) is innate, and 2. One’s ability to analyze one’s experience of the world is dependent on syntax (which is innate—UG), and NOT one’s particular language (content words).
However, I would argue that language tightly entwined with culture. One’s culture is something that is learned and essential for giving a word context and meaning. In other words, one’s particular language/culture affords new “subcategories” (borne out of the innate and universal categories), for which to understand and “do” in the world.
An fMRI study conducted by Gregory Berns et. al. in 2005 provides evidence to support Whorfian linguistic relativity. Their research shows how our perception can be warped by “what we are told about them”. It was found that group conformity is accompanied by functional changes in the brain. Therefore, what participants were told about their perceptions did not result in executive decision making, but rather, resulted in an altered perception of the world due to the social setting.
Berns, G. S., Chappelow, J., Zink, C. F., Pagnoni, G., Martin-Skurski, M. E., & Richards, J. (2005). Neurobiological correlates of social conformity and independence during mental rotation. Biological psychiatry, 58(3), 245-253.
http://www.ccnl.emory.edu/greg/Berns%20Conformity%20final%20printed.pdf
Manda:
Delete1. There is both innate and learned CP.
2. Universal grammar is innate; ordinary grammar is not.
3. I'm not sure what you mean by "analyze one's experience of the world," but if you mean analyze in words, yes, that requires syntax (of both kinds), but the meaning is in the semantics, not the syntax: grounded categories (content words).
4. "Culture" is habits and conventions shared by a population. But it's not your culture that dictates which mushrooms are edible and which not. It's nature.
5. Our thinking (cognition) is certainly influenced by what we are told. But CP is about perception and it is not yet clear whether language can alter perception. Category learning can; but category naming is not yet language.
Just to clarify, language ability is based on the innate and learned CP after which everything else is a combination of what is already learnt. What is the possibility for the evolution of language itself as a form of representation. Is there a combinatory stop after we exhaust every combination of words? Also if language is higher order cognitive function how come CP needs to compress because most categories that we use must have already been formed. Last, based on certain linguistic cues can we trace back the grounding process of how those cues formed?
DeleteI do believe that there are examples demonstrating how human perception of the world, not just our thinking, is influenced by what we are told. I understand ‘influencing perception’ in regards to the way a stimulus is realized by the senses and processed at the lowest-level in the brain, affecting the information that gets sent upwards for further processing and available for conscious thought. Weaker examples include: Evidence from music cognition research which demonstrates when musicians are led to believe the music they are about to listen to is more prestigious they report “liking the music” more and show greater physiological responses to acoustical cues in the music; The effects of neuro-enchantment, whereby participants rate explanations and reasons better in quality and more believable when extraneous neuroscience jargon is present in the observed text (the perception of quality is warped and the information that gets further processed is altered); Many cases of racism, especially in places that are racially homogenous, where implicit racial biases come from what a person has been told/ has learned rather than any direct experience (perception of threat is warped). A stronger example of altered perception that results from what a person is told is secondary traumatic stress, which describes symptoms like those of PTSD which an individual develops from hearing about the firsthand trauma experience of another. There are secondary trauma cases of therapists, lawyers, doctors…etc. who experience perceptual distortions like terrors, panic, and anxiety when confronted with stimuli related to the experience. What is perceived as threating or dangerous is altered at a very low-level, affecting the perception of an event and the immediate physiological response. In these examples language is influencing our very baseline level perception of the world. But, after just re-reading this response, I realize these examples apply more to continuous rather than categorical categories… In general, I believe that language can influence continuous perception more profoundly (or knowably at this point) than categorical perception – but I do think that language can create new categories. Like how the creation of new labels (especially those that driven by monetary incentives of big Pharma) for mental disorders has allowed people to be labeled and understood in new ways. As advertising and recognition of the labels grows, so do the proportion of people that get diagnosed, which may change the way people understand themselves and interact with and thus perceive the world. If these labels are understood categorically and are taken up in ways that don't necessarily reflect or get linked to an experience in a contrived way, they could be seen to create new categories of people in the way individuals identify themselves and others.
DeleteRe: “Once a set of category names has been ‘grounded’ through direct sensorimotor experience, they can be combined into Boolean combinations … which not only pick out the more abstract, higher-order categories … but also inherit their CP effects, as well as generating some of their own.”
ReplyDeleteWhile I understand the combination of grounded category names, I struggle with the idea that a higher-order category can generate its own categorical perception effects. In other words, how does something generate more than what it started with? I think that we approximate higher order combinations, but that the CP effects it has are solely inherited. I don’t see how exactly Bachelor adds a layer of separation/compression on its own independent of unmarried and man. It begs the question of where and how this added stuff comes from. The most prominent explanation that comes to mind would be if Boolean combinations somehow make something that is self-generative. If this is the case, then why are the Boolean combinations needed at all? Instead, I would argue that combinations don’t generate their own CP effects; rather, we approximate higher-order combinations by adding more and more combinations. An analogy would be a Venn diagram in which we must constantly add more circles in order to make a tighter approximation. This relates to our class discussion on why some students get the class content more than others at the end despite equal access to the lectures and readings (putting aside effort, attention etc.). It’s because higher-order combinations are complex in virtue of constantly having to make tighter approximations – i.e. adding more combinations. There is no self-generation; if certain ‘more basic’ combinations are missing, a higher-order combination can’t be reached. Furthermore, relating to the class discussion on apes, this can explain why a (presumed) lack of function words in apes explains their inability to make tight approximations/ higher order combinations. If this argument that it all depends on Boolean combinations and a rejection of self-generation, then I think the two problems mentioned in class might be able to make more sense.
Austin: I think you'll agree that whether or not categories learned from verbal definition (instruction) rather than sensorimotor induction can generate CP (compression/separation) is an empirical question, not one whose answer can be deduced by just thinking about it. In general, CP seems to be a way of making categories "pop out," by filtering out the irrelevant features and focusing on the invariant ones. It happens with our harder to learn categories, the ones that would benefit from more filtering, rather than the easier ones that pop out on their own (like mountains and valleys). When the category pop-out results from sensorimotor learning, the feature-detection and filtering is not based on known, named features; the feature-detectors are created by the learning process. With a category learned from a verbal description, I agree that it is more like the kind of "easy" sensorimotor category that already pops out naturally, like mountains and valleys, because the invariant features are already grounded and named. But what if the feature is a complex boolean rule: "Feature A and B or, if not C than not-A or D and..." Couldn't the automation of that rule with trial-and-error practice result in pop-out, in which the complex rule becomes a feature-detector too?
DeleteBut the only way to know is to test and see whether it generates CP.
I didn't say that apes just lack function words. (You can probably train those too.) What they lack is propositions.
So I still don't understand why they can't talk. (Nor why the same words don't have the same effect on a whole classful of students...)
I agree that it is an empirical question. Intuitively, however, my thought was that just as doing the symbol manipulation faster doesn’t address Searle’s Chinese Room Argument, so adding more combinations doesn’t seem to make anything more/new.
DeleteRight – my wording was incorrect. The thought was that function words are the hypothesized basis that apes lack for making propositions using categories. I would see trained “function words” as stimuli-response results and thereby, the apes wouldn’t actually be able to make Boolean phrases. My guess is that somehow function words are grounded for us yet not for apes. How? – I’m not sure.
It seems like the Whorf hypothesis originally proposed a strong idea of language determining cognition, but with evidence such as consistent color categories across languages, it seems highly unlikely. However, I found it interesting how it can still apply in terms of showing how language can influence or change perception in some respects, such as if human experiments could show similar data as the computer model that revealed categories acquired through language alone. But even if we found this to apply with people, how do we then show that this is an effect of language, not just naming? What kind of experiment would prove that language can change the way we think about something because of what we were told about it? Would these effects be specific to each language, or is going from naming effects to language effects just about going from more specific naming and categories to effects of general syntax and semantics of language/speaking?
ReplyDeleteJulie: Wouldn't CP induced by category learning through verbal induction rather than just sensorimotor induction (naming) be CP induced by language?
DeleteIn the UCSD talk, Lera Boroditski mentions how differences in vocabulary and different uses of words has an effect on cognition. First I was wondering if the effect was a correlation or a outcome, which might be possible to study with second languages. Second I started wondering if vocabulary had a stronger effect/correlation with cognition than syntax/grammar of different languages. The reason for my hypothesis was that I have heard of numerous examples of vocabulary differences between languages that are linked to a difference in cognition, for example the colour differences, or the object naming differences, etc. However, even after researching the topic, I can barely find anything on the links between syntactic differences and cognition. The examples I am referring to are the place of the adjectives (before the noun in english, after the noun in french), or the place of the verb (after the noun in french and english, at the end of the proposition or sentence in german). It seems to me like these structural differences are at least as important s the vocabulary differences to distinguish languages. Either the limited results is due to the fact that syntax is harder to study than vocabulary, or I would be very interested in knowing why one matters more than the other cognitively speaking. She did mention that a bilingual speaking russian will think more like a russian than whilst speaking english, which would lead me to believe that structure of the language matters just as much, does anyone have studied examples of that phenomenon?
ReplyDeleteJulie: I'm not sure whether syntax is less likely to influence cognition than vocabulary. But "cognition" is very general. CP is about changes in perception, and there would seem to be more scope for that with vocabulary -- but that also includes words whose meaning is learned by instruction rather than induction.
DeleteI've not yet heard of language-specific CP in bilinguals, but it would not surprise me -- especially in the obvious case of phonemes. In fact, in general, context may influence CP. Back to "information" again: in one context of alternatives, this is a chair; in another it's furniture; or mahogany; or "Chester," my grandmother's favorite chair... Each name (for the same referent) is based on a deifferent context of confusable alternatives (among which we need to reduce uncertainty), hence different features, a different feature filter... so why not different, context-dependent CP?
Once again, very insightful article. My first discussion point to everyone regards the question of whether categories and CP can be acquired through language alone. I think it is incredible that neural net simulations can emulate this process by grounded a category and then use “Boolean combinations” to create more abstract concepts of which direct experience is impossible. The article notes that this has not been demonstrated in human subjects. I wonder what experimental conditions would be sufficient to demonstrated this process in human subjects. Is the idea of a peek-a-boo unicorn not enough? Secondly, the article states “But of course our sensorimotor systems do not give equal weight to all features; they do not even detect all features. And among the features they do detect, some (such as shape and color) are more salient than others (such as spatial position and number of feathers). And not only are detected features finite and differentially weighted, but our memory for them is even more finite: We can see , while they are present, far more features than we can remember afterward”. This to me means that perhaps computer modelling must emulate this very limitation/restriction/finite nature. Interesting that something that is seemingly flawed, is integral and must be manually programmed into a device that may otherwise override these limitations.
ReplyDeleteI don't think categories can only be acquired through language. If a person is completely language free, they can still create categories. For example, if you pet a cat and feel that it's soft, even though you don't have the word for 'soft' you would put it in a different category than if you touched sand paper. Or, if you hear a loud noise and then no noise, you can categorize the noise as the existence of something. I do think that categories are somewhat based on at least one sense though.
DeleteThe term “categorical perception” (CP) refers to a critical process that allows us to make sense of the world around us. One result of CP is that we are able to label something we’ve perceived as either a member or a nonmember of a particular “categorical” category. Drawing from the reading, an example of a learned categorical category is “bird”. CP allows us to label penguins as members of the bird category, and Moorish idol fish as nonmembers. Penguins have important similarities to other members of the “bird” category (like having feathers and a beak) that Moorish idol fish do not. However, penguins are pretty different from every other member of the bird category in that they can’t fly. Further, penguins and Moorish idol fish share properties that birds do not: they’re both black and white with a little bit of yellow, they both lay eggs, and they both swim. We can say that CP has occurred here, because the differences between penguins and other birds seem smaller than the differences between penguins and Moorish idol fish relative to some baseline. In the case of the learned bird category, the baseline for comparison might just be the definition of a bird as a “warm-blooded egg-laying vertebrate distinguished by the possession of feathers, wings, and a beak”. This kind of baseline makes sense to me for a learned category like this one, where the important features for membership are definable. But for innate categories (especially sensorimotor continuous ones, like “red”), what is the baseline like? Do we have a prototypical shade of red in our mind that we compare perceived colors to, and use that to determine whether or not the perceived color is a member or nonmember?
ReplyDeleteIf you want to see a Moorish idol fish: https://www.google.com/search?q=moorish+idol+fish&espv=2&biw=1231&bih=607&source=lnms&tbm=isch&sa=X&ved=0ahUKEwipw4ibh5XSAhUj_4MKHdw1ATUQ_AUIBigB
DeleteThe motor theory of speech perception explained how speech was special and why speech-sounds are perceived categorically: sensory perception is mediated by motor production. Wherever production is categorical, perception will be categorical; where production is continuous, perception will be continuous.
ReplyDeleteI am a little skeptical about this. It is well known that there are differences regarding the obstruent perception depending on your first language. For instance, [p] in French has a similar VOT than [b] in English. Therefore, there is a point where the same stimulus will be perceived differently if you’re a native English or French speaker. However, I doubt that there are any differences in French & English speech motor execution. And if there were, then the point made in this article would be valid. However, if the two motor abilities are the same, then another process has to explain this discrepancy in speech perception.
Mouth motor abilities actually are different for different languages. Just like there is a critical period for language perception, there's also a critical period for language production. So, someone with English as a first language would have different motor abilities in terms of production than someone who grew up with French as a first language. For example, the sound th- is very common in English, but does not occur in French. From what I've seen, it's common for late-English learners to have motor difficulties in pronouncing th- sounds in English. There are many other examples of this kind. It can be trained and overcome (in some cases, but not always). Motor and perceptual abilities in terms of speech kind of go hand-in-hand. If you lose the ability to perceive a certain sound, you're most likely also losing the ability to produce that sound.
DeleteThe Whorf hypothesis as expressed here in terms of color perception does seem to be largely incorrect, but might it not still hold some ground in other aspects of language? Even if we agree that languages are universal in the sense that the Turing machine is universal - that is, that any language can express anything it is possible to express in any other language - our language might influence our cognition in other words.
ReplyDeleteFor instance, since languages developed in different environments, one may be more suited to express particular things than others (or at least in fewer words), and this could affect the scaling factor of categorical perception - the degree to which instances of the same category are similar and those of different categories are dissimilar. An imaginary language might have two separate words for 'large bird' and 'small bird', and thus native speakers could more readily distinguish between birds in either category than an English speaker with just the word 'bird.'
Languages also differ in the way they use gender and tense. In English only people (and some vehicles) have a gender, whereas in French all nouns are either masculine or feminine. Could these sorts of differences affect the way we perceive or cognize about people, places, objects, and time?
"Neural data provides correlates of CP and learning. Differences between event-related potentials recorded from the brain have been found to be correlated with differences in the perceived category of the stimulus viewed by the subject. Neural imaging studies have shown that these effects are localized and even lateralized to certain brain regions in subjects who have successfully learned the category, and are absent in subjects who have not."
ReplyDeleteI find this extremely interesting - categories seem to be localized in the brain! Of course this is all correlational - who is to say that it is the specific category causing the brain response rather than a third variable causing both the brain response and the categorical perception. What is it that scientists could do to solve this correlation problem?
RE: Learned CP
ReplyDelete"...recent demonstrations that although the primary color and speech categories are probably inborn, their boundaries can be modified or even lost as a result of learning, and weaker secondary boundaries can be generated by learning alone (Roberson et al. 2000)."
To me, this is the most practical and likely explanation and reconciliation of colour CP and the Whorf hypothesis.
Even if one was to argue definitively that some of these categorical perceptions are innate, formed by evolution and necessity, it seems that there must be room for learning and overwriting of these boundaries.
Intuitively that is the only way such CP's could form and evolve to meet our needs in the first place.
In my personal (and conveniently colour-themed) example, I have defined "tennis-ball" as a separate colour from both green and yellow, and would struggle to consistently place it "closer" to either one, as well as create a smooth gradient.
Funnily enough I am studying supervised/unsupervised, and behavior in neuroscience in another course right now. A bit higher up in the comment chain, Karl mentions that:
ReplyDelete"My understanding is that the higher “ungrounded” & abstract categories are actually grounded. Think of them like high up branches of a tree. The highest branches of the tree do not touch the ground themselves, but they are connected through a system to the roots that hold them up high in the sky, with the roots being the “lower” sensory-motor grounded categories. If there are no grounded roots, there could never be higher branches. "
This is a great way to phrase it. Is it somewhat like model based vs model free learning in behavioral psychology? Like the cached values vs inferred values of action and reward through the choice tree? Each stage of action and potential reward is assessed by the individual and action is assessed almost program-like?
I found 'the motor theory of speech perception' very interesting. I do agree that sounds like ba and da are distinct due to their differences in place of articulation (bilabial and alveolar). The distinction for ba and pa is based on voice-onset time (VOT), which is variable. Most occurrences of /p/ in natural speech will vary VOT. In fact, I took a class in phonetics that used a software (praat) to analyze my speech and for every instance of /p/ and /b/ I had differing VOTs based on the surrounding sounds. Therefore, I do think it is possible to produce sounds with a VOT in-between those of standard /p/ and /b/. I also think that our discrimination based on the sounds context should be considered. VOT will vary not only on the surrounding sounds, but the speaker as well. I am curious to see what the effects of context are on CP. Additionally, I feel that the Sapir-Whorf hypothesis is strengthened by sound segment discrimination. While infants can initially discriminate all sounds, they lose the ability to discriminate sounds in a foreign language with age, in this way it seems that language shapes the kinds of compressions and separations made. Lastly, I would like to make a comment about vowel separation. Martinet is known for having identified margins of security between vowels. This means that while vowels can vary in their height or backness of articulation within the articulatory space there has to be a distinction of space between which they are pronounced. It seems that this would support a categorical production distinction.
ReplyDelete"There are even recent demonstrations that although the primary color and speech categories are probably inborn, their boundaries can be modified or even lost as a result of learning, and weaker secondary boundaries can be generated by learning alone (Roberson et al. 2000)."
ReplyDeleteI'm having trouble wrapping my head around the idea of this. I think I could understand it more in regards to speech categories losing their boundaries through learning, as some languages have contrasting phonemes, whereas others may have two allophones for that same phoneme. Is this what that quote is getting at?
How would this apply to primary colors such that through learning their boundaries could be lost? Or is it simply just touching on the idea that through learning we are able to expand these inborn categories to include colors such as "scarlet" and "crimson" that may not be inborn?
Maya I don't think you can lose primary color boundaries; I think Roberson is talking about gaining secondary ones (like scarlet and crimson). But with phonemes, you can lose the innate ones, like r/l if your language does not use them.
DeleteIn response to "How the Languages We Speak Shape the Ways We Think"
ReplyDeleteI personally am very centered in my left/right distinction, as it applies to how I feel in my own body. Likely due to the fact that I am very left-side dominant, the two sides of my body do not feel equivalent to me, and I feel much stronger (not just physical strength) on the left side. I wonder how much of this would be effected if the left/right distinction weren't present in English? I have zero innate feelings about North/East/South/West--if I spoke the language mentioned in the video, would I perhaps feel more of that? It seems that since my left/right feelings are based on the fact that I am left-handed, which is rooted in the brain, not language, I would still feel similarly without left/right terms in my language. But I'm curious as to if there would be differences if my language never discussed those words.
Edit: perhaps "centered" was the wrong word to use in the first sentence....more like "aware" or "very NOT centered".
DeleteI am curious about that too! Since I am right-handed, I might not be able to share the same left/right feeling as yours. But I would like to share with you that, I found myself far more frequently using clock positions than left-right when I have to tell someone to look at a direction, and that makes me perform quite badly in answering questions that uses left-right to refer direction/locations (No matter I use English or any language). The languages that I speak all have the words "left" and "right", and they are more commonly used than other direction words. But interestingly, I personally stick to clock positions and end up I feel like, somehow, I have a clearer sense of where my object is in terms of clock positions than left-right. Since many of my friends can refer things by using left-right much better than I do, I guess language does play a big role in our spatial feelings, but what we keep repetitively practicing/using, might be more influential to us. (Just my personal view :)! )
DeleteDominique, I'm pretty sure that our perception of asymmetry and left/right is because of inborn asymmetries in our brains and bodies, but there is learning involved too, as even highly lateralized people (like you) need some practice to be able to say which is left and right.
DeleteAlison, I'm not sure what you mean by using clock positions instead of left and right: What position is the clock in, relative to your body? If you want to tell someone that something is on your right, do you say it's at 12:15 (assuming the clock's flat on the ground and you are standing in the middle, facing 12? But then how do you know it's not at 12:45 rather than 12:15 if you don't now left from right?
Children start out only being able to perceive directions egocentrically: relative to where I am, so, on my left, behind me, etc. When they get older they become able to perceive allocentrically: It's on your left. behind you. Different parts of the brain seem to be involved in egocentric and allocentric space perception.
https://en.wikipedia.org/wiki/Clock_position
DeleteThe clock positions I mentioned assume the person(subject) always facing 12. For example, if there is a baseball coming from the person's right-hand side then I will say: "There is a baseball coming from your 3-o-clock direction!", instead of saying his right. (However, in daily language, the clock positions 3,6,9 are less commonly used.) Since clock positions have 12 directions, I found clock position a more precise tool in referring directions.
In the section on Language-Induced CP, an idea is brought up that some of our categories must originate from another source than direct sensorimotor experience. For instance, we can have an idea of unicorns as being these mythological creatures that look like horses with a horn on their heads, but there has never been any evidence of unicorns actually existing. How do such ideas come to be? From abstracting to recognizing to establishing a name for something to symbol grounding to forming boolean combinations and to ultimately forming higher order combinations, we can come up with new categories that don't have anything to do with direct sensorimotor experience. It is said that higher order combinations pick out the more abstract, higher-order categories much the way the direct sensorimotor detectors do, but also inherit their CP effects and generate some of their own. This is an interesting discussion point since it hints at the idea of creativity. Given the basic building blocks that we have, how did we come up with language and with the capacity to form categories of things we have no direct contact with? Computational models such as back propagation nets can build up within-category compression and between-category separation as they learn. Is this similar to the causal mechanism responsible for categorization that operates in our brains? And with higher-order categories, it would seem reasonable to assume that there would be very little trial-and-error (i.e. for octacorns) so how can we keep such categories with minimal interactive experience?
ReplyDelete@Neil. If I understand your question correctly, you are wondering how we can abstract categories and create new categories (such as peek-a-boo unicorn) without any direct contact with the stimulus or object.
DeleteIt’s interesting that you link that with creativity as I think we can see different levels of abstraction as different levels of creativity. However, in this case, even the fact that we can form a simple sentence such as “It is cold today” is a creative act. Out of all the various stimuli I am presented with in this moment, I choose to recognize, identify, and communicate the coldness (which is just an amalgamation of many stimuli). If I was Funes, however, my teeth may be chattering, I may have goosebumps and my body may be stiff but I cannot understand all these factors as being cold, nor communicate them as coldness since I do not have an ability to abstract. To take perhaps an even “higher-order” category such as the unicorn. I have an image of a horse, another image of a horn which I had probably been exposed to at some point. Like chunking existing categories into a new category in the cold example, I’m doing the same with the unicorn. I guess this makes language itself a creative act, and our abstraction of even the most direct stimulus a creative act.
Neural plasticity seems to play a large role in what one is able to perceive and the precision with which a person manipulates an object. Development of these abilities seems to depend on the skill and attention paid to the senses and motor behaviours related to the task at hand - for example musicians develop an increased motor capacity (and associated coritical mass) for the fingers used to play their instruments. Lera Boroditsky presents the example of Australian Aboriginal languages giving their speakers an advanced sense of cardinal direction. This appears to be a setting where one needs to have an particular skill set (the ability to cardinally orient oneself) to communicate in a certain language. Instead of a skill or sense mediated by a need to play music, this increased perception is initiated and required by language.
ReplyDeleteIt feels like a bit of a late night, logical leap to move from sensorimotor plasticity and skill mediated by sensorimotor input to sensorimotor plasticity and skill mediated by language requirements. But, it seems reasonable that if language can require the development and change of categorical perception of sensorimotor skills relating to cardinal direction categories, then language can require the development and change the categorical perception of non-sensorimotor and abstract categories.
In regards to this section "though nativist theorists such as Fodor [1983] have sometimes seemed to suggest that all of our categories are inborn" – I'm wondering what the nativists use to support this claim? It seems like there is a lot more evidence towards some learning of categories and impossible for EVERYTHING to be innate. This is especially apparent from the evidence against nativism, "there are even recent demonstrations that although the primary color and speech categories are probably inborn, their boundaries can be modified or even lost as a result of learning, and weaker secondary boundaries can be generated by learning alone." According to this, then even innate categories are changed through learning, so nativism cannot explain all of categorization.
ReplyDeleteHave these people abandoned their views or do they have some other interpretation of this modern research?
I found human’s “categorical perception” really interesting. As in the example of varying “ba” and “pa” along a continuum, participants tend to perceive the sound jumps abruptly from one category to another. It is just like a square devided by two areas, on the left there is a darker area with its intensity gradually changed from light to darker and on the right there is a lighter area with its intensity gradually changed the same way. However, people tend to perceive a line in between (abrubt change) instead of gradual changes. I’m just wondering this categorical perception is formed by experience or some innate characteristics of human sensory system?
ReplyDeleteThere are studies that plays the sound ba and pa to infants and test if they suck (sucking is what infants do when they have arousal on a stimulus.). These studies were able to demonstrate that infants perceive speech categorically, as adults do. Since the infants being investigated haven't undergone phonological development(i.e. these infants were able to perceive virtually all the contrasts that human languages make use of), I guess categorical perception does not require learning of any language or experience.
DeletePeihong, CP for the ba/pa distinction is probably partly innate (chinchillas have it, and they never speak) and partly learned.
DeleteAlison, yes. pre-verbal children have the ba/pa CP before they speak, but probably hearing it (and then speaking it) sharpens the CP (and not hearing or using it would lose it, as in the r/l distinction in Chinese and Japanese"
So for those born with l/r distinction while losing it because of not using/learning it, does it mean that they can neither speak with nor hear the difference between the r/l sound? For example, for a Japanese person who does not distinguish between r and l, can s/he still tell the difference when somebody speaks English with an r/l distinction? That is, does it mean that s/he did not entirely lose the ability to tell the distinction, it’s just s/he has not exposed to learn to use/speak with the distinction?
DeleteRE: "The Whorf Hypothesis. We can now return both to the "Whorf Hypothesis" and the "weaker" CP for vowels: According to the Whorf Hypothesis (of which Lawrence's acquired similarity/distinctiveness effects would simply be a special case), colors are perceived categorically only because they happen to be named categorically: Our subdivisions of the spectrum are arbitrary, learned, and vary across cultures and languages."
ReplyDeleteI found the Whorf hypothesis highly intriguing and how it is highly linked to learned categorical perception. Since language is highly intertwined with how we perceive categorisation, are categories flexible in the sense that there is no real boundary? Words have meaning and that the meaning in the head is "grounded" to something vs. a word on a page is "ungrounded. However I feel like language works by consensus of the majority - when a word means something because enough people use it to mean that same thing. However, if people started using a word in a different way or ignored a particular rule or norm, then meanings can change. So judging by that logic, if we started saying that the colour blue is actually red, would we be able to change our categorisations based off of "how we sort and name them"? Although we categorise through trial and error, when exactly does 'wrong' become 'right'? The line seems to be very blurry.
I don't think the actual name you give to a category is as important as the invariant features found in members of that category. I'm not sure changing what we call light of the wavelengths 450-495nm (blue) to another word (red) affects the category in any significant way. If it did, then the categories for colors in every language would be different (as they generally have different vocabulary). Also, how would this work with someone who is bilingual?
Delete“The nets accomplish this by selectively detecting (after much trial and error, guided by error-correcting feedback) the invariant features that are shared by the members of the same category and that reliably distinguish them from members of different categories; the nets learn to ignore all other variation as irrelevant to the categorization.“
ReplyDeleteIf the nets are detecting ‘invariant’ features while ignoring any variation that is irrelevant to the category, I am wondering where the idea of context comes in. As mentioned in the earlier paragraph 'Categories: categorical and continuous', “There are range or context effects as well”. So for categories that are continuous at the microscopic level, such as color, how would a neural net accomplish accurate categorization overtime if variation, such as context for example, that is irrelevant to one situation is relevant in another?
I think refining what you mean by context would be helpful here. If you define context as generating a specific output when given a certain range of inputs, then I could see a neural net in response to certain data giving a different output. I don't understand what you mean by context being irrelevant in one situation and relevant in another. Do you mean that certain features of information are more salient depending on what is being categorized? For example, if when encountering two colour pigments C1 and C2, the back propagation learning algorithm picks out certain visual features of the stimulus to place each into it's own distinct category, thus when it encounters a scene with C1 and C2, it is able to pick out each colour. But when presented with a new stimulus C3, C1 and C2 both appear to be the same categorical colour, then in the context of C3, C1 and C2 are the same. A learning algorithm could do this by adjusting the weights in such a way that when given the 3 as input, a different output is produced.
DeleteThis reading made me curious whether all Categorical Perceptions are innate, or different between individuals/cultures. For example, colour CP is innate, and does not differ between individuals, but in Japanese ‘l’ and ‘r’ are not distinguished (if I remember correctly). This was mentioned in class to be a loss over time, that young infants in these cultures can hear the difference but adults cannot. The section titled ‘Learned CP’ suggests that this sort of learned modification to CP is common. If this is the case, can we ever separate learned CP from innate CP? Are there any which are purely one or the other, or is CP always a combination of innate structures and learned differentiation criteria?
ReplyDeleteYoung infants who haven't undergone phonological development are able to perceive virtually all the contrasts that human languages make use of. I guess learned CP is a loss of some of the innate CP ability after infants have exposed to their mother language.
DeleteI found it interesting to read that perception of phoneme categories is present in chinchillas and infants who cannot yet speak. This makes sense to me given the studies I have learned about, related to continuous perception in infants. For instance, in one study, very young infants were not able to predict that, if a rolling a medium-sized cylinder caused an object to move a certain amount, a large cylinder should cause it to move more. This study appears to show that, before the age of 5.5-6.5 months, children are not able to apply continuous perception of size to the world around them, even though they can (presumably) still perceive differences in the size of the cylinders. Therefore, there must be a similar step in categorical perception during which infants have the ability to perceive different phonemes but are not yet able to understand how to apply this perception to the objects and interactions they face in the world. In the case of chinchillas, they will never be able to.
ReplyDelete“...to show that it is a full-blown language effect, and not merely a vocabulary effect, it will have to be shown that our perception of the world can also be warped, not just by how things are named but by what we are told about them.”
ReplyDeleteI am still puzzled by the definition of a category. An apple belongs to the category of fruits. The same apple can be used as a placeholder or even a weapon if throw down a skyscraper. The varying functions of an apple allows it to be “categorized” as a placeholder or a weapon. If function is not the determinant of category, then are we attempting to understand the fundamental category in which the apple belongs to regardless of its varying functions?
Diane Ackerman’s book One Hundred Names for Love demonstrates this adequately:
What can you do with a pencil--other than write? I’d begun. “Play the drums. Conduct an orchestra. Cast spells. Ball yarns. Use as a compass hand. Play pick-up sticks. Rest one eyebrow on it. Fasten a shawl. Secure hair atop the head. Use as the mast for a Lilliputian’s sailboat. Play darts. Make a sundial. Spin vertically on flint to spark a fire. Combine with a thong to create a slingshot. Ignite and use as a taper. Test the depth of oil. Clean a pipe. Stir paint. Work a Ouija board. Gouge an aqueduct in the sand. Roll out dough for a pie crust. Herd balls of loose mercury. Use as the fulcrum for a spinning top. Squeegee a window. Provide a perch for your parrot….Pass the pencil-baton to you...”
“Use as a spar in a model airplane,” Paul had continued. “Measure distances. Puncture a balloon. Use as a flagpole. Roll a necktie around. Tamp the gunpowder into a pint-size musket. Test bon-bons for contents...Crumble, and use the lead as a poison.”
Some of our categories must originate from another source than direct sensorimotor experience, and here we return to language and the Whorf Hypothesis: Can categories, and their accompanying CP, be acquired through language alone?
An item that belongs to one category can easily change into a different category which is what continuous perception describes, whereas categorical perception explains the sameness of a category and the distinctiveness of a category. Perhaps, categories are self-similar (i.e., a category has the same items as one or more of the categories). This fractal-like may be the reason for the phenomenon of CP--although a category shares similarities with another, it is still unique in itself, which is another property of self-similar objects many overlook. The difference in the sameness is what may need to be address for the problem of “why the same words don’t have the same effect on a whole class full of students” to be resolved and consequently the Whorf Hypothesis.
Re: A category, or kind, is a set of things.
ReplyDeleteDo categories necessarily have respective words to to represent them? I concern experience which has no existing word/descriptor for it, like how “a picture is worth a thousand words” suggests an element of lived experience not entirely translatable into language. These experiences if considered as categories would be an unnamable aesthetic causing a sense of spontaneity/awe. Are anomalous experiences which de/re-construct existing categories not categories themselves?
RE: Categorical Perception
ReplyDeleteCan some categories only be acquired via linguistic induction, or is it possible for every word to be learned instructively via error-corrective feedback (in the style of “this is a bachelor, that is not”)?
At least at the time this paper was written, no direct human behavioral evidence has been found for language-induced CP. Has this changed since then? Have computational models provided any evidence for this effect?
RE: CP and Symbol Grounding
Why can’t organisms as smart as apes have the capacity for language? Can apes understand numbers? If so, that would demonstrate some capacity for abstraction. Do apes have the capacity to pick out every word in our 1500-word minimal grounding set? If they do, is the ability to apply logical operations to this grounding set *the* defining capacity that sets apart linguistic from nonlinguistic animals?
What percentage of the 1500 are innate categories?
Regarding the part about Berlin & Kay (1969) showed that there is no Whorfian learning effect with colors:
ReplyDeletehttp://languagelog.ldc.upenn.edu/nll/?p=17970
The link is an article about a colour perception research on the Himba tribe in Namibia where the habitants' language no word for blue nor distinction between blue and green. The research found that the habitants were not able to distinguish the blue colour from a bunch of very green colour (which is obvious to us), but could be able to distinguish a green from a bunch of green colours (which look all the same to most people). (There are pictures inside if anyone wants to have a look!) I couldn't find the exact video recorded, but I have seen the video that showed they really did a very amazing job in distinguishing the greens. I found this could be somewhat consistent with the Whorf's hypothesis.
I guess what Berline & Kay showed suggests that language may not be the ONLY factor that controls our categorical perception, they may also showed us how strong language and learning could be, in modifying our categorical boundaries.
Somehow, I wonder if colour categorical perception is not just a result of learning different colours, but also a loss of ability to see and distinguish ambiguous colours. If Himba habitants are having such ability to see the special green from a bunch of greens, do we also have such an ability when we were infants?
Alison, Alison, I'm not really up to date on the latest in color perception, but if it's really true that (1) the Himba cannot discriminate blue from green and (2) can discriminate greens better than us, then that would be a Whorfian effect. (I find (2) more believable than (1), but who knows?) I have never heard it suggested that (primary) color discrimination could be lost from disuse, but that may be because unlike phonemes, which one could really never hear or use in some languages, all (primary) colors are everywhere. (Even the Himba have the sky and water, if they don't have blueberries.)
DeleteBut if these are Whorfian ones, they are weak ones, compared to the standard rainbow.
I also read about the OvaHimba and their enhanced ability to discriminate between shades of green. I wonder if any of this could be attributed to their outdoor lifestyle as semi-nomadic hunters and gathers rather than linguistic categories. Maybe a portion of the compression/separation of colour categories innate to us all is enhanced in the OvaHimba as a result of their extensive exposure to/interaction with things that are green in nature. This hypothesis could be tested by looking for similar patterns of colour perception among other hunter-gatherer societies. Also, I read somewhere that the colour blue was the last colour to have emerged in many ancient languages (don’t know how true this is, but if this is really the case, then the pattern observed here seems to lend support to the Whorf hypothesis).
DeleteRE: Liberman et al. (1957) reported that when people listen to sounds that vary along the voicing continuum, they hear only ba's and pa 's, nothing in between. This effect -- in which a perceived quality jumps abruptly from one category to another at a certain point along a continuum, instead of changing gradually -- he dubbed "categorical perception" (CP)
ReplyDeleteLiberman’s observation is very interesting in that the majority of human sensory perception is categorical. Is it advantageous that our sensory perception is more adept at perceiving a change in something rather than a continuum? For example, our visual system is tuned to pick up sharp and contrasting changes in spatial or spectral information. A sudden change in luminance prompts more response than a gradual change. (See the Craik-O’brien-Cornsweet illusion). In addition, the hierarchy of the visual system encodes more complex “categories” of information at higher levels of processing. Could this be evidence of similar compression and separation of visual information across cultures?
I’m curious as why the Motor Theory of Speech Perception was abandoned, since I think there is some merit to it. The idea that our sensory perception is mediated by motor production seems viable and offers a logical explanation for the linguistic CP effect; that, when we vary speech sounds along a continuum, listeners can only perceive two discrete sounds (/ba/ or /da/), but no intermediate sound. Since our motor apparatus developed such that it cannot produce the intermediate sound, our phonetic categories formed accordingly.
ReplyDeleteThe Motor Theory got me thinking about how mirror neurons work, and how the neurons that are active while we execute a certain action are also active when we see another do that same action. Perhaps in the same way that our mirror neurons recognize a motor action as a certain category of movement (i.e. seeing someone wave their arm activates mirror neurons that are active when we, ourselves, wave our arm), our phonetic categories, and the sounds we recognize, are implicitly formed by our ability to produce those sounds.
The section on “Acquired Distinctiveness” does, however, provide evidence that refutes the theory’s claim that “the CP effect is a result of learning to produce speech”, in that infants and chinchillas exhibit speech CP, even though neither can speak. However, that it is an innate effect evolved to “prepare” us to learn to speak does seem to make some sense – especially, since we are born with both a universal speech production and perception ability, and depending on the sounds we are exposed to during a critical period, it molds our motor ability and determines which sounds we can and cannot produce. I’m not sure if this train of thought entirely makes sense, but my point is that our perceptual categories do seem, in some way, to be grounded in our motor ability, leading me to suggest that there is some truth to the Motor Theory – perhaps it shouldn’t be totally abandoned, but rather revised accordingly.
RE: Motor theory of speech perception
ReplyDeleteWhile I understand this theory has been discredited I wonder if there is something to be said about the contribution of biological elements to the innate categories discussed - I'm thinking more specifically about the McGurk effect where there is a perceptual confound between visual and auditory information. What is there to be said about the way all senses integrate basic sensory information, such as in the OT of the brain where visual and auditory information overlap and the fact that this is mapped in a spatial manner onto the brain contribute to the limits of what we perceive as categories.
Maybe I'm totally off-base but it would be interesting to understanding if and where the category boundaries may be drawn in the human brain.
From the article, CP can be transferred through language. From the video, different language result in different CP. However, it is true that this difference in perception is not static. A person is capable of learning new languages and therefore capable of experiencing new perceptions. Further, the ability to translate between languages proves that a humans perception is not bound to her language. Language is a tool to aid in perception, but language is not perception. Humans are bound to their sensory modalities, that is, for example, sound frequency between 20 Hz to 20 kHz, light in the visible spectrum, somatosensation above and bellow certain degrees, movement bellow certain speeds, and so on. In other words, human perception is constrained their body. ALL humans are bound to the SAME dimension. This means that ALL humans are capable of perceiving the same things. Language is NOT an aspect of this dimension, what we are capable of perceive is not bound to the words/categories we have. However, what we perceive can be influenced by words/categories.
ReplyDeleteDoes this argument coincide with your notions? I think I am slightly confused.
I think that maybe your point runs parallel to the Linguistic Relativity Hypothesis that we talked about earlier: The way we view the world can be manipulated by our language. I do think it can occur but I also believe that human are capable of deconstructing what we’ve learnt in order to change our perceptions. If we can initially build up categories and kinds, why wouldn’t we be able to change them around with learning?
DeleteI find it interesting how our anatomical features, or the way our body works, feedback onto the way that we cognize and categorize. In the example with the vowels, pa/ba use the same mouth movements and therefore are more similar and harder to distinguish than pa/da. I don’t know if this is maybe a scenario of which came first the chicken or the egg, but I’m wondering if these categories are separate because of our anatomy, or that because the categories are separate our mouths and lips evolved to be different. It would be interesting to see how this would play out with someone who maybe had some mouth deformation, and whose pronunciation of pa and ba involved different mouth movements. Would this create separate categories?
ReplyDeleteI guess the anatomy problem can be explain by the concept affordance, our anatomy only allows us to pronounce certain sounds and not the others.
DeleteI think that is an interesting point, our anatomy but also our brain chemistry and biology, afford us the ability to make these particular interactions with the world. To what extent does the connectivity and anatomy of our brain afford us the ability to cognize? This brings up the degrees of freedom involved in reverse engineering a T4, though we all have slightly different anatomical differences and probably connectivities, we all can cognize. Primate brains are similar to humans, and yet their doesn't afford them the ability to be conscious. The limitations of what we can and cannot do are set by our anatomy/biology, and yet we seem to be able to do much more beyond. I think it's interesting to think about the limits of our cognition, but also how limitless consciousness appears relative.
Delete
ReplyDelete"These language-induced CP-effects remain to be directly demonstrated in human subjects; so far only learned and innate sensorimotor CP have been demonstrated (Pevtzow & Harnad 1997; Livingston et al. 1998). The latter shows the Whorfian power of naming and categorization, in warping our perception of the world…. to show that it is a full-blown language effect, and not merely a vocabulary effect, it will have to be shown that our perception of the world can also be warped, not just by how things are named but by what we are told about them."
Although this last point is difficult to prove empirically, it seems intuitive to me that the Whorfian hypothesis could be taken further, as categorical learning is influenced not only by the vocab colloquially associated with the thing, but also what we are taught about the thing. I think that the first step in categorical learning is that the emotional tone and images that are associated with the new word at first experience are parcelled together in episodic memories. For example, I think that whenever most Western children first encounter the word “evil”, the category primed consists of vivid images of Disney characters from recent movies they’ve seen. At this first stage, they can’t use the word fluently. They aren’t able to use the word in speech and so simultaneously cannot apply the concept to other instances in their life. I don’t think this new category is useful until it is primed in more contexts. Once the category is primed in a sufficient amount of contexts, a more durable, efficient, and more abstract category pertaining to the word takes shape, and the child can appropriately use the word in speech.
I think language goes a step further than simply helping us create categories. Language allows us to recognize which referents may be innate vs. which ones are learned. For example, everyone can agree on what blue is, what is hot, or what is cold. In contradiction, cross cultural notions of beauty, or wealth may vary and have different referents.
DeleteSymbols like ‘blue’ have meaning. It appears that in every culture, the names or symbols referring to specific colours have been grounded to the same referents. Blue IS blue in all cultures. There are no cultures that have used representative symbols (names of colours) to refer to the wavelength interval ~ 490–450 nm that is different from what a native English speaker calls ‘blue.’ In French or Spanish ‘bleu’ or ‘azul’ MEAN the same thing as blue.
The reading highlights that speech sound ba-pa are innate in humans, and it appears that so is the perception of referents such as colour. In other words, the Whorfian hypothesis that states that our cultures change the way we perceive the world. But this does not seem to apply here. This contradiction to the Whorfian hypothesis can be seen as evidence that colours are FELT identically across cultures. This supports claims of language as a mind reading miracle. Cross culturally, symbols have been generated that have the same meaning regardless of the cultural background of that person.
Reading this paper confirms my original thought that our cognition is indeed influenced by what we are told and the language effect; however, not in total accordance to the Sapir-Whorf hypothesis. As Sapir-Whorf hypothesis mainly states that two languages may be so different from each other as to make their native speakers experience of the world qualitatively different from each other. As the paper have already pointed out, that this is simply not the case. Whorf’s view on colour terms and Inuit snow vocabulary are later unfounded, by researches demonstrated by Berlin (1969) for colour spectrum and both Martin (1986) and Pullum (1991) for Inuit snow terms.
ReplyDeleteWe know that categorical perception (CP) is the relative abrupt perceptual change at the boundary with distinctive feature of a compression and separation effect. Furthermore, there are evidences, which showed the boundaries of the inborn categories such as the primary colours and speech could be modified as a result of learning; and more interestingly, also through learning, a weaker secondary boundary can be generated. What is fascinates me is the newly learned, weaker secondary boundary. Drawing from the paper’s section about the “weaker” CP effects for vowel, it seems that it defined “weak” CP effects as its production is continuous rather than categorical, but its perception, by the criterion of their context, is categorical. Continuing to extrapolate, does that mean that learning categorization to induce CP with the audio stimuli and the timbre multidimensional sounds will result in a new, secondary, weaker boundary? And if we amplify the duration of learning and time, it seems like this new boundary can eventually modify the original boundaries to finer grains.
To loop back to the Whorf Hypothesis and its relation to categorization, the paper poses near the end “can categories, and their accompanying CP, be acquired through language alone?” In class, we have discussed that an alternative way to learn category is with narrate and grounded words, which is demonstrated by the “bachelor” example in the paper. So following this, if it is possible to language-induce CP effects in human subjects (which I believe so), then it will confirm that our perception of the world can be warped by not just how things are named but by what we are told about them. So in trying to map the experiment we have in lab, a lot of our texture properties and perhaps even our acoustic stimuli is hard to verbalize and most of the learners are not explicit – so I am thinking of ways, of how we can narrate in words our properties to these subjects so as to create a language-induce CP effects…
DeleteGrace I think you raise an interesting point. To provide some anecdotal information, one of my Spanish teachers made an interesting observation on the culpability of her Spanish students (from Spain) and her American students, when they forgot their homework. To say I forgot something in Spanish, it would translate to "my homework forgot itself". She said this translated into the students caring much less, or feeling much less guilty when forgetting their homework because the blame was essentially on the homework itself. Again this is purely anecdotal.
DeleteBoth innate and learned categorical perception are sensorimotor, though the paper seems to suggest that language plays a large part in this, that sensorimotor experiences are combined with names that have been grounded, which allows for higher-order combinations meaning that these words can be used to categorize more things by combining words.
ReplyDeleteIf I allude to a scenario I mentioned in past comments and consider a single person, and that person were to be living in this world completely alone with only nature (so without other human beings), this categorization would still take place but not be attached to any symbols (words), they would be attached to mental images. The process of combining thus becomes more difficult, and probably would not occur without a lot of mental strain (if it were to come about at all, because communication doesn't occur). We wouldn’t be able to more efficiently use what we learn, and to then learn more without difficulty. This appears to suggest that language is potentially one of, if not the most important aspect in the development of cognition. Consciousness would not be affected but without language, how far would cognition slump?
RE: The motor theory of speech perception
ReplyDeleteI’m not quite sure if I understand how this paper is distinguishing whether speech-sounds are innate or learned. From what I’ve read it seems as though “categorical perception” is when the categories are clearly distinct and do not “change gradually.” Since ba and pa are distinct it means that they are categorical perception. In the reading for 6A it says that categories are tied to learning so does that mean that it is not innate for speech sounds?
RE: The innateness of sounds
ReplyDeleteI think proof that at least some of the sounds infants make are innate lies in nursery forms. According to this, nasals [m,n,etc] are associated with females due to nursing. The physical act is similar to mouthing those phonemes; hence, "mama". This similarity is spread through many languages. On the other hand, 'pa' and 'ba' are associated with males, like "papa".
While this could be due to distant genetic relation of languages, it could also be that these sounds are innate to humans. Also, this could be the first forming of categories of speech sounds and somewhat of an origin point. Nasals are 'female' and other types of stops are 'male'. If this theory holds, that would mean that the one phoneme is grounded to a meaning for that infant.
I was very interested when reading about the Wharf Hypothesis, because I saw a really strong parallel with what Lera Boroditsky was saying about how the languages that we speak shape the way we think.
ReplyDeleteIf the words that we use in our everyday vocabulary affect our world view, then the languages that we speak affect the categories that we create in order to orient ourselves, as well as our memories and our perception. However, what does this mean when thinking about reverse-engineering cognition? If we successfully build a T3 that speaks English, and then we want to build one that speaks German, will we be able to follow the same guidelines?
In the same vein, I was interested in the video about how our lexical categories can affect our sense of orientation/geolocation. It strange to reflect on how our spatial relations are relative to an arbitrary frame of reference. Would a speaker acquiring English who didn't have the same spatial relations as we do necessarily have to learn how to re-categorize his/her existing spatial relations, grounding them indirectly from verbal feedback and correction, or would he/she be able to carry on just fine in our linguistic society by making reference to his/her native categories?
DeleteI wonder how Liberman's study asked the participants to place the ba's and pa's on the voicing continuum – I can definitely see how a categorising bias would arise if the sounds were played and participants told to put them into the ba or pa category, but I'm fairly certain that a study structured this way would defeat the purpose. I assume that the sounds were simply played for the participant, who was asked to respond what they heard. Assuming what Harnad says in this paper about our physiological inability to actually produce the sounds between ba and pa, I wonder if that is what causes the participants to categorise those in-between sounds so rigidly – they simply cannot produce the sounds in between. I'm curious as to what the results might look like if the participant was asked to respond on a continuous visual analogue scale, like that used in pain research (a simple line with no other markings or gradations, and the participant is asked to draw a line on the spectrum).
ReplyDeleteI would think that learned CP would occur in language in addition to, for example, the ba/pa and ba/da effect. For instance, in English there is a clear distinction between the L and the R – Would an English speaker exhibit CP for L and R that a speaker of a language that didn’t have the L/R distinction wouldn’t exhibit? I wonder if this has been tested. If so, it seems like this would be a clear example of learned CP in language.
ReplyDeleteIt's also interesting to speculate how much, if at all, this influences our perception. We may say two things are different categories, but do we actually PERCEIVE them differently? We've all had arguments about whether something is purple or pink, for example. Do you and the person who claims the sweater is purple actually perceive it in a different way?
The idea that something may be perceived along a continuum (varying degrees of possessing that feature) yet still illicit the perception of a distinct category rather than varying degrees of membership is something that is so interesting. CP has been found in speech with the categorical perception of ‘pa’ and ‘ba’ depending on voice onset time and in vision where the perception of either ‘ka’ or ‘ga’ depends on the mouth movements created rather than the sound recorded. Is CP necessary for passing T3 or cognizing or is it merely a by-product of our hardware?
ReplyDeleteThe idea that something may be perceived along a continuum (varying degrees of possessing that feature) yet still illicit the perception of a distinct category rather than varying degrees of membership is something that is so interesting. CP has been found in speech with the categorical perception of ‘pa’ and ‘ba’ depending on voice onset time and in vision where the perception of either ‘ka’ or ‘ga’ depends on the mouth movements created rather than the sound recorded. Is CP necessary for passing T3 or cognizing or is it merely a by-product of our hardware?
I wonder if categorical perception and learning extend to all optical illusions, as well language and colors. Much like how those involved in the ‘ba and da’ study mention how the phoneme jumps from one category to another, there exist optical illusions, like the spinning ballerina, where an image is spinning in one direction, and then just by focusing on a small portion of the ballerina she starts to twirl in the opposite way. In others, there just seems to be a mass of dots when, by focusing on the negative space or by some certain twitch of our muscles we see a dalmatian, and cannot unsee this dalmatian ever. We can go back to seeing the ballerina any which way we want, but the dalmatian remains a dalmatian no matter how much we convince ourselves it’s just a bunch of dots. Does anyone have any ideas on why we can see can see the original ballerina, but not the original dots?
ReplyDeleteI found it interesting how the definition of categorical perception has changed over time. The beginning of this article clearly distinguishes continuity from categorization, classifying categorization as something that is strictly all-or-none. But it was then discovered that CP is not so 100% or 0% after all. The contemporary definition being “CP occurs whenever perceived within-category differences are compressed and/or between category differences are separated”; this shows how even categorical perception is somewhat of a continuous scale, with boundaries between categories being less or more blurry depending on the category. This is referred to later as the “accordion effect.” My question is, how do we still distinguish between categorization and continuity; is continuity to categorization also a continuous spectrum?
ReplyDeleteRE: “…Humans have such innate category-detectors too: The human face itself is probably an example.”
ReplyDeleteThere is a debate in literature about whether face perception in humans is domain general or domain specific (For those interested: Kanwisher, N. (2000). Domain-specificity in face perception. Nature Neuroscience, 3, 759-763 AND Tarr, M., J, & Gauthier, I. (2000). FFA: A flexible fusiform area for subordinate-level visual processing automatized by expertise. Nature Neuroscience, 3, 764-769). The domain generality side of the argument claims that humans’ unique ability in face recognition is mainly the result of high frequency of their exposure to faces and that they develop an expertise in face recognition due to the importance of faces as stimuli. I’m wondering whether innateness of the category of faces would go against the domain generality of face perception argument. Alternatively, could it be the case that our ability to detect faces as a general category is innate but eventually we develop the ability to recognize them and associate their features with people around us which would be an instance of learned categorization.