Cognition: Human vs Non-human Animal

It seems to me you are skirting the issue with a definitional approach. Please define primitive concept. :-)

Good comeback. :-) Okay. Piaget coined pre-concept for about ages 2-7. He didn't coin an analogous term for the sensori-motor stage before about age 2. So I will define primitive concepts as the infant's manner of thought as it gradually adapts motor behaviour to sensory input to serve its motivational goals, increasing the number and complexity of its sensory and motor abilities.
http://www.encyclopedia.com/doc/1O87-preconcept.html
http://www.encyclopedia.com/doc/1O87-preoperationalstage.html
http://www.encyclopedia.com/doc/1O87-sensorimotorstage.html

A non-understanding memorization is at the root of what I call rote learning. It's how we teach dogs tricks.

What's the basis of memory, if not recording and recalling similar or identical events or things?

Post 41

Friday, October 3, 2008 - 3:35pm

Hi Ed,

In a way, you could say that the dog has generalized ("I get an instance of the group of yummy things when I obey"), but all the dog did was memorize and repeat -- with the added incentive of pleasure and pain. There's no logical categorization, just sentient-driven memorization.

How did the dog form that mental group of yummy things? That is key . . . and cannot be explained via association. But I invite you to try.

Jordan

Post 42

Friday, October 3, 2008 - 3:57pm

Jordan,

Maybe I'm not grasping some complexity here, but it seems easy to make the case for the association of yummy things.

With a computer program I'd just make a link from the 'percept' of a dog treat to the stored sensation of hunger being satisfied - a pleasure. It is a simple many-to-one link because the 'percept' of a tidbit of steak, the 'percept' of a piece of chicken, etc. - all just link to that pleasure of satisfying the hunger. There is no advanced cogitation involved in determining via logic what the CCD is. The link is automatic. Little floor-vacuuming robots could have programming like this to 'learn' what is furniture and change their routes through a room. There is no 'concept' of yummy treat - just another association of hunger to multiple pleasure-from-satisfying-hunger remembered events. A dog MUST generalize percepts, not concepts, to some degree or all of reality is new every instance and no learning of any kind could ever occur - no action would be possible.

Post 43

Friday, October 3, 2008 - 5:00pm

Hi Steve,

Well, the simple sit scenario does *not* indicate whether a dog forms or can form a concept of yummy things. I was just going along with Ed who suggested that we could say that the dog formed a category of yummy things.

It's a little off-point, but to explain the problem with the sit scenario, the dog associates the word "sit" with the action of sitting, which in turn is associated with something good for the dog, i.e., a treat. There's no indication that the dog has formed a yummy-things category here. Being given a treat (and eating it) does not indicate a selectivity behavior, which is what we'd need for identifying a mental categorizing process.

SO for example, if a dog recognizes the word "sit" coming from different voices -- as evinced if the dog sits on their command (and doesn't sit when given some other sound) -- then that would indicate mental categorization, for the dog would have to isolate the sound into a specific category, regardless of voice.

Incidentally, "generalized percept" doesn't make sense to me. A percept is one perceived concrete object, a singular. Generalizing a singular doesn't make sense to me.

Jordan

Post 44

Friday, October 3, 2008 - 5:58pm

Hi Jordan,

Let me talk about "generalized percept" first. That may be a bad choice of words. I recognize that the existent is singular - one single unit. Let's say we are talking about my neighbor, Tom. One single, unique person. I am able to "generalize" the sense data as part of the automated process of converting sense data into a percept. He could be wearing clothes that are unusual for him that I have never seen, a different haircut, and yet I perceive Tom. The ability to lump together differing, discrete sensory impressions is similar, as a process, to conceptualizing in omitting aspects until the identification is made and then subsuming those aspects as new properties of the percept (Tom has wrap-around sunglasses now). For a human, we can exit the perceptual process and go into logic to resolve some cognitive dissonance. If I see someone that looks like Tom but isn't, I might have to actively reason it out - can't be Tom, he is out of town, and he is taller then that guy. Or, all of a sudden realizing that the person in front of me is Tom, but I hadn't recognized him, because I am on vacation in a different part of the country and he is a very unexpected part of that context. I wouldn't be surprised if a dog also had confusion when it gets mixed signals - it's eyes saw you drop what looked like food in its bowl, but it doesn't smell like food... Maybe the dog tastes it and that determines if it is added to the set of individual units that the dog will eat (fits the dog's category of food-by-association - "by association" since it doesn't have a concept of the same kind as a human).

Look at the example of the dog sitting. There were always dogs in our house as I grew up and I know that many dogs will respond to the command "sit" from someone they have never met before. And that they may also respond to other commands. They have somehow "generalized" the sensory packet result enough to here the pattern of the word even if there are some variations in the frequency, timber, volume, inflection, etc. Now, if you say that "selectivity" of some kind is required to constitute a category, then isn't this a category? Again, I see that it is a category but that it is formed at a perceptual level with nothing more than association.

What am I not getting here?
(Edited by Steve Wolfer on 10/03, 6:00pm)

Post 45

Friday, October 3, 2008 - 6:02pm

Much of this so-called confusion seems to stem from doing what is "in effect in mid-stream", instead of grasping it from the initial and simple first stage primitive organisms use [the 'pleasure/pain' association avoidance] and moving along the line, showing the stages as they come, and how in these more complicated stages nothing in the manner of any cognitive attributes take place - until one reaches the human stage... understanding how these more primitive constructs operate leads much to understanding where the 'break-off' point is in percepts and concepts - in cluding those 'pre-concepts' being mentioned - and why, because without grasping the why, there is no understanding of the necessity of the 'coming into being' that takes place...

Post 46

Saturday, October 4, 2008 - 8:31am

Jordan,

Next, I think the accusation of anthropomorphism is misplaced for at least two reasons: (1) the accusation is typically made on unempirical grounds based on the notion that there is a categorical difference between the cognitive traits of humans and animals. I try to start from empirics and see who passes the tests. (2) the accusation begs the question as to whether animals share some cognitive traits with humans.

Heeeyyy. You can't use (1) and (2) because they violate the principle of parsimony. Due to the large difference between animal and human behavior, the simplest assumption explaining that difference -- the proper null hypothesis in any scientific investigation into the matter -- is to start with the notion of different cognitive traits (and see if it can be disproven empirically). It's not proper to start with the notion that animal cognition is the same as human cognition -- and then look to see if you can prove different. It's proper to start with the notion it's not the same -- and then see if you can prove different.

But anyway, I maintain that concepts are subjective to the end that essences change per an individual's personal experience. That is, people add, grow, and shrink their mental categories.

I have to concede this, but check this out:

Let's say you think justice is essentially retributive, rather than reformative to the violator, crudely restitutive to the victim, or deterrent to the population of potential violators.

Let's say I think establishing justice involves cases where the punishment doesn't or shouldn't fit the crime -- i.e., that the purpose of punishment is not to right wrongs, but to either reform violators (where the best justice means punishment too light to fit the crime) or that it means to deter others (where the best justice means punishment too harsh for the crime). If necessary, just pick one of these two non-retributive views of justice and continue to follow what I'm saying.

The shared feature between us is that wrong actions require response. That consequences themselves are moral. We start with that. We wittle it down. We solicit each other to integrate our view of justice with other views until ... one of us is more right than the other. Whoa! How did we achieve that?! Holy Kahmohlee!

And we will always -- by pressing each other for non-contradictory integration -- achieve that kind of agreement.

If an advanced alien civilation wanted to make us lab rats, and removed all conceptual awareness from living humans then, in a few millenia (assuming we survive), we would be right back where we started with the same concepts for the same things. It's what Rand said when she said that conceptualization, while performed by man, was dictated by reality. The aliens can't change reality (nobody can), they can only remove our heretofore accumulated conceptual understanding of reality.

But, because concepts are objective -- we'd get right back to where we were before (with enough time). We might have different sounding names for things, we might call the concept of love "googenschnootzle" or we might call republicans "socialists" ... [scratch that last] ... but we'd have the same concepts for the same things -- because concept-formation is dictated by reality (which would be the same because it always is).

Ed

Ed

Post 47

Saturday, October 4, 2008 - 9:19am

Very good point, Rev'.

Ed

Post 48

Saturday, October 4, 2008 - 9:31am

Jordan,

I was just going along with Ed who suggested that we could say that the dog formed a category of yummy things.

You're putting words in my mouth -- like dog treats. Give me more credit than you would to 'man's best friend,' okay?

:-)

I said dogs get vague groupings based on pleasure-pain sentiment and rote learning (non-understanding memorization).

We humans can talk about these vague groupings as if they were clear categories in the animals' minds -- because we, not animals, have the mental ability to talk about clear categories. We can even list out the things that would fall into those vague groups. For instance, we can say that a piece of steak would fall into the category. We can say that a dog treat would, too. We can also say that a handful of nails -- while initially appealing to the dog because of the remembrance that past things given have always been good tasting -- decisively wouldn't end up in that "category."

Ed

(Edited by Ed Thompson on 10/04, 1:40pm)

Post 49

Saturday, October 4, 2008 - 9:49am

Steve,

... many dogs will respond to the command "sit" from someone they have never met before. And that they may also respond to other commands. They have somehow "generalized" the sensory packet result enough to here the pattern of the word even if there are some variations in the frequency, timber, volume, inflection, etc.

Now, if you say that "selectivity" of some kind is required to constitute a category, then isn't this a category? Again, I see that it is a category but that it is formed at a perceptual level with nothing more than association.

Right. I've had dogs, too. If they've rote-learned the commands enough, then even a toddler -- a being smaller than them, with a super-mousy voice -- can tell them to sit and they will. They think the toddler will give them a treat. Little do they know that, if they can't reach the treats up on the counter-top, the beings littler than they can't either. After several, several obey-on-commands, the dog will give up sitting at the never-ending command of a tireless (and treat-less) toddler. They begin to learn a new association -- that the kid is full of shit.

:-)

They become "selective" in who they listen to, in response to the associated memory of who is giving them treats. Now, at some point, there will be children tall enough to reach the dog treats on the counter-top, and the dog will get a chance to form a new association of a member of the family with treat-giving. Us humans know that the differentiating factor for the category of treat-givers is body height (actually: arm reach). But the dogs just associate and memorize the different particular entities that are treat-givers -- remembering those who aren't (without understanding why one gives treats but others don't, just rote learning who does).

In essence, it's behavior that makes it look like a category was formed, but no category was required to explain the behavior (just remembered associations).

Ed

Post 50

Saturday, October 4, 2008 - 8:30pm

Hi Ed,

the simplest assumption . . . is to start with the notion of different cognitive traits. . .

I'm just starting with a testable theory, and those who pass the test win, regardless of their species. With this approach, I don't need to compare species. If the subject can pick which item over here belongs in the group over there, and if the subject can pick which item does not belong in the group here, then the subject wins; that is, the subject can form concepts.

Per your bit on justice. No disagreement. Just way to abstract for our current sandbox!

. . .we would be right back where we started

The fact the concepts change over the course of one's life, over the course of human history, and even over the river, through the woods, and in a different country suggests the contrary.

You're putting words in my mouth -- like dog treats. Give me more credit than you would to 'man's best friend,' okay?

Credit? In this market? Psha! But seriously. I'm sorry. When I said you'd suggested that dog's formed concepts of yummy things, I was referring to your statement --

"I get an instance of the group of yummy things when I obey

-- and I took "group" as synonymous with "category," so I thought it okay to say to Steve that you suggested we could say dog's formed categories of yummy things. You hadn't said "vague groupings" then, so I figured my bit to Steve was fair. But okay. Onward . . .

Steve,

Your example of the ever-changing Tom is an example of a percept. Objectivists could've viewed each instance of Tom, regardless of his outfit, as a member of the concept TOM. I'd be fine with that, but I don't think that's how Objectivists frame the situation. Objectivists require two or more objects to serve as members subsumed in a concept.

. . . I know that many dogs will respond to the command "sit" from someone they have never met before. And that they may also respond to other commands. They have somehow "generalized" the sensory packet result enough to here the pattern of the word even if there are some variations in the frequency, timber, volume, inflection, etc. Now, if you say that "selectivity" of some kind is required to constitute a category, then isn't this a category? Again, I see that it is a category but that it is formed at a perceptual level with nothing more than association.

To answer your question: yes, it's a category, and it is a category of percepts, of various instances of the sound "sit" with the variance in their measurements omitted. Wouldn't you agree? And isn't this just Rand's basic definition of concept?

Also, it is not a category formed at the perceptual level. Categories, even very basic categories, are always abstracted, hence always above the perceptual. Objectivists view the perceptual level as being an immediate and automatic awareness of concretes. The category of "sit" is categorically (ha!) different from the concrete "sit."

Honestly, though, the sit scenario is not the best test for determining concept formation. I prefer tests where I can physically see the subject sorting it all out -- like have the subject remove from a pile of things the thing that doesn't belong, or have the subject put into a pile of things the thing that does belong. You know - group stuff.  In the sit scenario we are stuck inferring that when the dog sits, it's because the dog mentally sorted out the "sit" sound. Could've been a false positive.

Jordan

Post 51

Saturday, October 4, 2008 - 9:19pm

Jordan,

I'm just starting with a testable theory, and those who pass the test win, regardless of their species. With this approach, I don't need to compare species. If the subject can pick which item over here belongs in the group over there, and if the subject can pick which item does not belong in the group here, then the subject wins; that is, the subject can form concepts.

Does it matter if the groups are memorable to the subject (as in requiring only sense perception, crude association, and a working memory of all the particulars)?

It's important when researching animal cognition to account for the confounding variable of differences in memory. For instance, there's the Clark's Nutcracker that stores pine nuts in 5000 different places in the fall in order to get ready for the winter -- and remembers where most of them are!

Now, I couldn't remember 5000 particulars like that. I could "remember" 5000 particular places if I had went by an abstract math formula, such as burying my nuts ... no wait, that didn't sound right ... such as burying dollar bills in 5000 different places, but always exactly 10' away from each other.

Science has to prove that particulars weren't memorized before the idea of animal concepts is entertained as a viable explanation for animal behavior. With the Clark's Nutcracker, that may mean using more than 5000 bare particulars to test for category-formation.

Ed

(Edited by Ed Thompson on 10/04, 9:20pm)

Post 52

Saturday, October 4, 2008 - 9:57pm

Jordan,

The fact the concepts change over the course of one's life, over the course of human history, and even over the river, through the woods, and in a different country suggests the contrary.

That's not a fact! Concepts never change, only peoples' background knowledge does -- which affords for subdivision of broad concepts or integration of narrow ones. The concepts themselves -- now viewed as objective, immutable units -- don't change. Here's Rand:

An objective definition, valid for all men, is one that designates the essential distinguishing characteristic(s) and genus of the existents subsumed under a given concept�according to all the relevant knowledge available at that stage of mankind�s development.

All definitions are contextual, and a primitive definition does not contradict a more advanced one: the latter merely expands the former.

Definitions preserve, not the chronological order in which a given man may have learned concepts, but the logical order of their hierarchical interdependence.

Recap:
The chronological order of concepts learned is of little significance, but the logical (read: objective) order of their hierarchical interdependence is significant. Objective definitions of concepts are valid for all men, everywhere (given a level of knowledge).

Here's more from ole' LP:

The fact that certain characteristics are, at a given time, unknown to man, does not indicate that these characteristics are excluded from the entity�or from the concept. A is A; existents are what they are, independent of the state of human knowledge; and a concept means the existents which it integrates. Thus, a concept subsumes and includes all the characteristics of its referents, known and not-yet-known.

Recap:
A child's concept of table or dog, is the same concept of table or dog of an adult. The concept doesn't change through time -- it always refers to the same referents. Only our knowledge changes through time, not individual concepts.

And finally:

This does not mean that conceptualization is a subjective process or that the content of concepts depends on an individual�s subjective (i.e., arbitrary) choice. The only issue open to an individual�s choice in this matter is how much knowledge he will seek to acquire and, consequently, what conceptual complexity he will be able to reach. But so long as and to the extent that his mind deals with concepts (as distinguished from memorized sounds and floating abstractions), the content of his concepts is determined and dictated by the cognitive content of his mind, i.e., by his grasp of the facts of reality.

Recap:
Concept-formation is an objective ("same for all") process. All concepts are the same for everyone. The only difference is in how much complexity you seek to acquire in your conceptual "library." Think of concepts as if they were playing cards (Ace-through-King). They're the same playing cards no matter who holds them, but players with more knowledge can hold more cards in their hands (or know more about how they interrelate).

Ed

Post 53

Saturday, October 4, 2008 - 10:45pm

Jordon,

According to Rand, my neighbor, Tom, is a concept. You are right that ordinarily it takes two or more units to make a concept, but proper nouns are an exception. "Every word we use (with the exception of proper nouns) is a symbol that denotes a concept, i.e., that stands for an unlimited number of concretes of certain kind. (Proper nouns are are used to identify and include particular entities in a conceptual mode of cognition...)" (ITOE, pg 10) However, she appears to contradict herself on page 175 where she says, "...a sound, if it is to be a word, can not denote objects directly, without representing a concept. (A word which did that would be a proper noun.)"

Clearly, I could not be aware of Tom without first integrating packets of sense data to make a percept. I am assuming that Rand's position is that I convert that percept into a concept by attaching the name "Tom." The only units subsumed are past, present and future instances of Tom.

At this point, I wonder about "unit" as an idea since when I identify any unit, like "this table" - which would only make sense when we had a context that established the table referred to by "this," I automatically subsume past, present, and future instances of "this table" which require a concept and that leaves percepts as strictly sense data organized and only as far as existents, but not given any identity. I might see a table, but to refer to it, makes it "this table."

But putting aside the issue of proper noun concepts, the fact remains that identity is attached to what comes in as sense data that varies from instance to instance (Tom in a suit, Tom with sunglasses, etc.) There is some form of generalizing or massaging of the sense data going on to arrive at the identification.

You said, "Objectivists view the perceptual level as being an immediate and automatic awareness of concretes. The category of "sit" is categorically (ha!) different from the concrete "sit."

Because the audio transmission of 'sit' can vary in clarity of pronunciation from identifiable to unidentifiable and because there is a range of 'sit' vocalizations that will be recognized, there has to be some kind of generalizing routine in the dog's mind (or a human's) that converts that audio sense data packet into the percept 'sit.' That would be granting implicit existent status to the sense data, That makes it possible to move to the next step where it can be recognized (associated with yummy treat and the sitting motion) - it has a primitive form of implied identity - a primitive concept but only done by association.

Post 54

Sunday, October 5, 2008 - 6:36am

Adler considered them as 'perceptual abstracts'... The Difference of Man and the Difference It Makes

Post 55

Sunday, October 5, 2008 - 7:41am

Steve,

I'm in agreement with Jordan regarding the treatment of proper nouns as non-conceptual. However, proper nouns are open to the workings of memory and association.

To a non-human animal, every entity is a proper noun.

Ed

Post 56

Sunday, October 5, 2008 - 10:04am

Hi Ed,

It would be a problem with testing if the subject were simply remembering each particular (buried nut) and responding the same way to it (digging up and eating it). But this problem is controlled in tests that have the subject deal with novel, as-of-then-not-memorized or even known, items.

Concepts never change. . .

My point was that individuals and groups of people, in different eras and areas, have different sets of concepts. I think that's a fact. And there's no way we'd retrieve those sets perfectly were we forced to do it over again. It's aking to perfectly re-forming the complete works of Shakespeare from scratch.

And . . . Rand quote suggests that concepts do change. Concepts change because when definitions change, concepts grow or shrink (non-contradictorily, if you like); that is, they include or exclude more referrents than they did before. In this sence, concepts are not the same for everyone. The child's concept of "table" is not the same as the adult's concept of "table" -- at least not until they include the exact same set of referrents. And what's more, the "table" can stand as a member of many different concepts, not all of which are necessarily shared.

Steve,

Let's avoid the proper noun oddity by simply swapping Tom out for that tree over there.

the fact remains that identity is attached to what comes in as sense data that varies from instance to instance

I agree. Objectivists call that identification a percept.

there has to be some kind of generalizing routine in the dog's mind (or a human's) that converts that audio sense data packet into the percept 'sit.'

You say generalizing; I say concept forming. :-) But I need to change one of my conclusions a little, but farther, I think, from yours. (poop!) Here's why: 'Sit' differs from other percepts, specifically percepts that are objects, because the sound of 'sit' doesn't just sit there (ha!) like trees and table do. It flits into and out of existence. Each instance of 'sit' is a different percept, whereas each instance of that tree or that table is the same percept.

The difference between 'sit' and those objects is just one more reason the sit scenario is bad for testing concept formation. Given this view, my new conclusion is now that the subject has formed a concept it lumps all instances of 'sit' into the same category, even if each instance of 'sit' comes just from one voice. It's important, so I'll ask again: Doesn't the subject form a category of 'sit' when it lumps together various instances of the sound of "sit" with the variance in their measurements omitted? And isn't this just Rand's basic definition of concept?

Cheers,
Jordan

Post 57

Sunday, October 5, 2008 - 11:17am

Ed wrote:

Due to the large difference between animal and human behavior, the simplest assumption explaining that difference -- the proper null hypothesis in any scientific investigation into the matter -- is to start with the notion of different cognitive traits (and see if it can be disproven empirically). It's not proper to start with the notion that animal cognition is the same as human cognition -- and then look to see if you can prove different. It's proper to start with the notion it's not the same -- and then see if you can prove different.

How about 'similar' rather than 'same'? What about evolutionary history and similarity of neural processes?

Maybe I need to read some Adler, though.

(Edited by Merlin Jetton on 10/05, 11:19am)

Post 58

Sunday, October 5, 2008 - 12:15pm

Jordan, Ed,

I have to think a bit on this issue of proper nouns.

Jordan, you say, "Doesn't the subject form a category of 'sit' when it lumps together various instances of the sound of "sit" with the variance in their measurements omitted? And isn't this just Rand's basic definition of concept?"

Here is what I see happening when someone tells the dog to sit:
1) The audio and visual sense data packets arrives at Fido. Sense data philosophically implies existence - but not as objects - not yet.
2) The sense data packets are parsed automatically and the sound 'sit' is isolated from background sounds, and the vision of the person is isolated from background objects. This completes the conversion from sensory data to initial percept. Now there is an implied existence and that which exists does so as objects. This step is necessary to put short term memory to work. A tape recorder picks up all undifferentiated sound with no distinction between foreground and background. We have to isolate them to store them - I'm assuming a dog does the same.
3) Some routine is able to label the combination of short duration sound from a human who is looking at Fido while speaking as a possible command. An association previously established this and it is hooked to positive motivation.
4) A search routine goes to work to see if this sound is a known command. The routine somehow compares this particular 'sit' to the set of learned commands to find similarities (certain ranges of frequency, duration, volume, etc.) - automated. If this routine finds a match, there is an implied identity. Philosophically all things that exist must exist as something - have some identity, but not necessarily as identified objects.
5) A match is found and that association formed in the past between that command and the motion of sitting is carried out. The identification makes that 'sit' one that is subsumed under the concept 'command' - but that is for humans. To use the word 'concept' with a dog as if it were the same thing is an error.

Even though a dog never forms a concept for existence or identity, they still are required to understand the dog's epistemology - the dog must have a sense of existence to put routines to work - even if it is automatic. And the concept of identity is driving identificiation which is the heart of association.

So, Jordan, as to your question. No, it is not a category in the human being's use of the word. The particular measurements are not omitted. Instead it is a fuzzy search - one that says, find something that is similar to these measurements. Look for a stored command to associate with, one that has a short duration, one with an abrupt ending sound and ssss starting sound. We humans can see that volume could be omitted, but it might not be for Fido because the dog doesn't conceptualize CCD like we do - it might instead just do a fuzzy logic search on all sound attributes - meaning it might not respond to a yelled or a whispered 'sit.' Instead it is more likely just branch-tree logic used in a fuzzy search routine to find a match. 'Sit,' 'stay,' and 'fetch' could be examined to see if they matched the audio sense data packet - actually the short term memory of the audio packet. It would be going down the list of commands because of the association of being looked at by a human, who is nearby, who utters a short sound and keep on looking at Fido. That would be the associated trigger to search the command list.

A human could hear a command, go through the same set of routines, and let's say find no match. The human could decide that it was intended as a command, and imagine ways that it could not be recognized, "Is this guy speaking English?" "Does he have a speech defect?" Then the human could imagine different resolutions to the scenario - creating a picture of what the next few minutes might look like if he did x, or if he did y, and then he could choose which scenario to put into play depending on what he decided best suited his goal in this context. It is that leap into imagination and that evaluation of goals and the choosing from imagined scenarios based upon values that is the uniquely human form of awareness.

Post 59

Sunday, October 5, 2008 - 1:32pm