It was that time of the year again for me, time to attend the latest (10th) Quinn Memorial at UBC and to write and reflect on the issues raised. This time around we had the pleasure of seeing Dr. Susan Carey, a Harvard professor, talk about concepts. The title of the talk was “Concept Acquisition: Beyond Logical Construction and the Building Blocks Model.”
Susan Carey was introduced by UBC's Head of the Psychology Department Geoff Hall who enumerated all her distinguished awards and accomplishments and summarized her view as one that gave more credit to infants' minds than Piaget had done previously. In fact, her views were also opposing a number of ideas propagated by Locke and Berkeley.
All this sounded interesting and aroused my curiosity. I have often felt that Piaget had generally underestimated the rich and resourceful mind and the mental and other capabilities of children, but it would be much better to actually hear it from someone who was an expert on the matter.
Yet as so often happens, I was disappointed at first. What she was talking about mostly had little to do with what I thought she was going to talk about. In fact, it seemed initially that she was not showing us how children are smarter than we think, but that they, in fact, deceive us!
But first thing first. Susan Carey asked us the simple but poignant question of why understanding can be at times easy and at other times hard. The general view is that we are born with a set of innate primitives. This is basically our knowledge base that can increase its content but not its processing capacity. In other words, we are operating with an 18-month processor.
According to this view, our learning cannot increase our expressive power. Put differently, we are rather limited in terms of learning and understanding new primitives since we have already acquired the necessary linguistic and semantic blueprint, a set that is somewhat set in stone. But Susan Carey disagrees with this view since new primitives can be learned.
She gave us an example of certain migrating birds. They travel over long distances and do so at night. How do they know where to go? Is it based on a set of innate primitives or do they learn and adjust? Or in that specific case, how did the birds know where to go in the dark?
One theory is that they may have used the North Star Polaris. But how did they know which one is the right one to follow as following an erroneous star could take you - or rather the birds - to the wrong place? Also, what is the North Star for us now, has not always been so due to the Earth's rotational axis; in fact, about 14,000 years ago, it used to be the star called Vega (and it will become Vega again in 12,000 years or so).
This cannot be information passed on genetically from bird to bird generations. There must be some learning involved, that is the ability to create new primitives. That is when the computational primitives come in. This is not just using your processor, but also making it more powerful through the power of arithmetic.
How does this knowledge happen and does it apply to humans as well or is it simply for the birds? There are two methods we apply to learn about numbers. One of them is the Parallel Individuation Model. This means that we learn and count each number at a time, and see each number as distinct and separate from other numbers.
Yet there is also a process called the Analog Magnitude Model. In this case, we process chunks of information at once and see them more as a comprehensive set rather than as individually different or distinct items. The ability to do this changes with practice, experience, and age, but as a general rule of thumb, we can pay attention to and “hold” 3 or 4 items at a time.
Susan Carey then presented us with a bunch of dots grouped together and asked us to guess how many there were. For lower numbers where less crowding occurred, say 7 or 8 dots, we could make more confident and accurate guesses, but once there were twenty or thirty dots, there was too much noise and distortion, and we would be often wrong in our estimates.
Hence, she was explaining the acquisition of concepts via a mathematical/computational manner. I felt a bit disappointed because I had been more interested in concept-making in terms of language and its representation. Nonetheless, there were interesting bits of information that caught my immediate eye and attention. For instance, there was the surprising fact that children learn numbers at an early age, but they do not “understand” them! In other words, they can count from 1 to 10, but they do not know what that means!
She showed us some videos of experiments done with young children. When they were told to give a certain number of toys and they had awareness of that number, they would do so correctly. However, if they had no knowledge of that number, they would err. For example, a child that does not know anything beyond 3 would grab an indiscriminate amount of toys. They could still “count” up to ten, but did not notice that the number “4” corresponded with the four items in front of them, that is 4 toys put together.
This was very interesting as we often show off the knowledge of our children without awareness of the fact that their counting and these numbers had no tangible relation with the facts and abilities! One child, for example, would comment “Daddy, Mommy, and me” to talk about any items in a set of 3. This shows that she has awareness that the set of 3 corresponds with three, in this case very specific, items.
In a similar way, according to embodied simulation, this is how we learn our first language. We have an image in our head and the spoken or written word is used as an analogy; they are paired and associated with each other using a representational scheme. For example, the word One would be associated with “finger” and that would then lead to a long-term memory of that particular concept, i.e. number.
We often learn concepts and use logic to connect them with others, hence building connections within our minds. But not all learning processes as we have seen are through logic alone. We often use mathematical representations. We know that adding one more to any set increases the number and value of the set by one.
It may take us a few years to be able to accomplish this feat, but at a certain age, we understand it. This then can be expanded and applied to a number of other computations, hence growing and diversifying our capacity to learn. In other words, she has shown us that learning increases over time and is not limited to a set of primitives.
Now if you are slightly confused, you are not alone. As I am wont to do at such events, I looked to personally chat with our presenter Susan Carey for some clarifications. With my red wine in hand, I approached her at the reception and asked her about embodied simulation and Piaget. She gave me an answer that clarified my doubts and confusion.
According to her, embodied simulation is correct but a too simplistic view and account of human learning. We are capable of much more. When a person sees a dog, they do not simply associate the animal with the word “dog,” but concept building goes beyond that. The person makes a wide range of assumptions, such as the fact that there are many of its kind and that this particular animal is different from other animals, say a tiger or a squirrel.
Some of these assumptions may be wrong or mistaken, but they are still part of the inner world that the individual carries around with him or her. We can see that this is not just associating one thing with another à la Piaget, but that children at an early age already make a number of assumptions vis-à-vis what they see. This shows more activity and awareness of the human mind than was previously assumed, and it may not be necessarily limited to humans as animals seem to draw conclusions and notice connections as well.
All this left me inspired. There is more to human learning than meets the eye. I imagined the brain being capable of doing an infinite number of tasks like the endless possible moves on a limited chess board. As this was going through my head and our conversation had reached its end, she surprised me with the following question: Was I a computer scientist?
I admitted I was not. I do not think that watching Mr. Robot would make me a computer expert and my general suspicion regarding technology has always prevented me from embracing technology more than was necessary or pragmatic.
It was also the first time I had been associated with computer science. Perhaps it was due to my question, which she deemed both relevant and appropriate, or perhaps my look (I think I was wearing a hoodie). Be it as it may, I left this talk feeling slightly more accomplished knowing that I had added to and updated my knowledge base.