How I became a heretic

by Neil Rickert

It was somewhere around 1988.  For various reasons, I became interested in trying to understand learning.  I already new from my own experience at growing up, that human children can be excellent learners.  And my experience as an educator (university professor) supported this view.

I also knew, as a practicing computer scientist, that machine learning did not work at all well.  The kind of machine learning that worked best was reinforcement learning.  But the difficulty was you had to give a direction to the learning system, and come up with a reward system for the reinforcement.  So it was hard to judge how much of the learning was due to the programmer, rather than to the software.

My starting assumptions

When I started this project, I did not expect to succeed.  I knew it was a difficult problem.  I did better than I had expected.  And that is probably because of my starting assumptions.  However, my starting assumptions were apparently quite different from those of epistemology (the branch of philosophy that studies knowledge).  So I guess my starting assumptions were the start of my philosophical heresy.

What was obvious, was that some children grow up in places where cats and dogs are commonplace.  Others grow up in places where kangaroos and koalas are more common.  And still others grow up in places where you are more likely to see polar bears and seals.

In short, we are a diverse species.  Children grow up in a wide variety of environments.  And they seem to do well and learn well in any of those environments.

It seemed very unlikely that knowledge of all of those environments could be in our genes.  The size of the genome seems too small to encode that much.  So the alternative was that, instead of being born with innate knowledge of the environment, it must be that we are born with innate abilities to discover what kind of environment we are in and to discover how to cope with that environment.

And I guess that made me some kind of empiricist.

Empiricism

There’s an long standing argument within philosophy, between empiricism and rationalism.  According to the rationalists, we already have a lot of innate knowledge at birth.  According to the empiricists, we are not born with innate knowledge, but instead acquire that knowledge by learning.

Looked at, in those broad terms, I was some kind of empiricist.

When I later looked at the literature on empiricism, it didn’t fit.  John Locke, one of the founders of British empiricism, did seem to be talking about acquiring concepts.  But he did not provide a persuasive account of how those concepts are acquired.

When I go past Locke, and look at Hume, then his empiricism is mostly about acquiring beliefs.  And that idea of empiricism as acquiring beliefs seems to be the main theme of empiricism.  I don’t see how that could work.  As I’ve already mentioned, my view of the problem faced by a child, is that the child must form suitable concepts for the environment in which he finds himself.  And it is that acquiring of concepts that is at the core of the learning that we see in children.  Acquiring beliefs, once you already have the concepts, is a simpler problem.  But, by itself, belief acquisition cannot account for a child’s learning.

So I was already a heretic at that point.

The tabula rasa

In later readings, I came upon this, from Margaret Boden (in “Computer Models of Mind: Computational approaches in theoretical psychology”):

Presumably the child brain is something like a notebook as one buys it from the stationer’s. Rather little mechanism and lots of blank sheets. …

Our hope is that there is so little mechanism in the child brain that something like it can be easily programmed.

Boden is criticizing the idea of a tabula rasa.

It must be that my understanding of tabular rasa is very different from Boden’s.  I have taken it to mean that the child comes knowing nothing about his world.  But that requires that the child comes with a lot of innate abilities to find ways of making sense of the world.  Boden seems to think that the tabula rasa implies no knowledge of the world and no innate abilities.

Apparently my ideas on the problem of learning are very different from what philosophers see as the problem of knowledge.

Ontology

Philosophers study ontology separately, as part of metaphysics.  They seem to take ontology as a starting point.  The see ontology as a prerequisite to acquiring knowledge.  Ontology is what you can talk about, and knowledge is taken to be the beliefs that are the content of your knowledge.

With my way of looking at learning, this does not make sense to me.  What we talk about is something that is up to us to decide.  It isn’t given as a prerequisite to having knowledge.  If anything, knowing what we want to talk about is a central part of our knowledge.  And it doesn’t really matter what exists.  We can talk about unicorns, even if they don’t exist.  As a mathematician, I can talk about numbers, and whether numbers actually exist does not seem important.

Summary

So there you see some of my heresies.  What I see as empiricism is very different from what is discussed in the literature.  How I see the tabula rasa appears to be different from the way that philosophers and psychologists see it.  And I don’t see the point of separating out ontology from the problem of knowledge.

Advertisements
Tags:

7 Comments to “How I became a heretic”

  1. I used to follow feral children research years ago, skills like language and empathy must be learned fairly early or the ability goes away. I’m not up on current status of all that, but both of you are probably partially right.

    Liked by 1 person

    • Yes, I have followed that. But my reading is a little different.

      As I see it, the feral children build a system on concepts of their own, because they are not part of a society. When they are brought into society, they find it difficult to adapt to the society’s system of concepts, because that would require them to abandon their own conceptual system. That is to say, the problem is the conflict between their own conceptualization and the conceptualization within the society that adopted them.

      Liked by 1 person

  2. Interesting post Neil!

    “It seemed very unlikely that knowledge of all of those environments could be in our genes. The size of the genome seems too small to encode that much. So the alternative was that, instead of being born with innate knowledge of the environment, it must be that we are born with innate abilities to discover what kind of environment we are in and to discover how to cope with that environment.”

    I remember talking about this to some degree in a previous post of mine:
    https://lagevondissen.wordpress.com/2015/06/13/neurological-configuration-the-prospects-of-an-innate-ontology/

    “The see ontology as a prerequisite to acquiring knowledge. Ontology is what you can talk about, and knowledge is taken to be the beliefs that are the content of your knowledge.”

    I think one could say that ontology is a subset of one’s knowledge and is a prerequisite to having any OTHER kind of knowledge, for example, any beliefs based on those fundamental concepts. I see ontology as a kind of fundamental epistemological “grammar” needed before you can make any sense or use of any kind of epistemological “language” (causal relations and the beliefs that are derived from them). In order to “know” something, based on my personal definition of knowledge which is roughly “causal patterns recognized that allow one to predict the future, which can therefore be used to successfully accomplish goals”, one has to have a way of organizing those causal patterns into certain objects or sets with particular relationships between those objects or sets of perceived/conceived phenomena. I would say that this organizational schema is one’s ontology (or epistemological “grammar”).

    “What we talk about is something that is up to us to decide. It isn’t given as a prerequisite to having knowledge.”

    I think however, that one’s ontology is still needed in order to talk about anything at all (without it being utter, incomprehensible nonsense), whether those things actually exist or not (e.g. horses, unicorns, numbers, etc). We don’t have to have a well-grounded ontology that we’ve explicitly formulated like some arm-chair philosopher, but we do have to have an ontology implicit in our thinking, some kind of conceptual framework, how we see the world as structured, etc. And this means that our ontology limits what we can and can’t talk about, even imaginary things, because those things can only be comprehensible if they are conceived of in a way that can be handled by our ontology. I can imagine a unicorn, because for all practical purposes it is just an object with a particular shape and various qualities. Whatever ontology is needed for a horse will likely do for the unicorn as well, but if I had no ontology (even implicitly in my brain) that would support the existence of objects, or perhaps animate objects, then I couldn’t even imagine any kind of animate object, let alone something as specific as a unicorn. I think that’s what those philosophers mean when they say that ontology is “what we can talk about”. It limits the possibilities (even if it’s an infinite set, it’s smaller than other possible infinite sets), and it limits them even if within those possibilities we can choose to talk about any particular logically conceivable option. My two cents anyway.

    Like

%d bloggers like this: