Thoughts // Patterns – Part 1

Some people might object to the notion that computers can have thoughts. Indeed, some did so in the wake of my previous post. The popular argument went along the lines of “humans can have original thoughts, computers only do what they’re programmed to do”. For the sake of further discussion, let’s examine the premises of this argument.

The first is that humans have original thoughts, which I suppose implies free will. But let’s start with the second premise: that computers don’t have ‘original’ thoughts, only predetermined processing of input. One could respond “go ask someone whose computer just crashed what input they used to make it do that”. But even if we chalk such occurrences up to thermodynamics. it’s not as easy to dismiss some of the seemingly creative products of AI development (like the image shown below). Of course you could argue that this “AI” is still a program, only with enough layers of complexity to produce unpredictable results.

Computer-generated image on the theme 'pagodas'.

Image generated algorithmically from random noise, on the theme ‘pagodas’ (by Google).

Which naturally brings us back to the first premise, the originality of human thoughts. I don’t have anything original to add to the fundamental debate of free will vs. determinism, so let’s limit ourselves to a weaker question: Can we confidently say that our conscious thoughts do not arise from a very complex ‘program’?

One objection from the field of biology is that we haven’t yet learned to read the instructions of this ‘program’ we’re born with, so how could we say what might arise from it? Sure we’ve sequenced the human genome, but until we know the function and regulation of each gene (plus the epigenetics and protein interactions) all we have is a text, without punctuation, in a language where we don’t have much grammar or vocabulary.

Another confounding fact is that a neuroscientist can put you in a scanner and tell you things that you didn’t know were happening in your brain. Heck, a trained poker player can do that. And if we’re only partially aware of our own thought processes, how well-founded can a subjective notion of having original, voluntary thoughts be? It doesn’t seem like such un-thought thoughts are confined to background processing. One can easily point to instances where we act with little guidance from our conscious thoughts, or where we have thoughts which we can’t control (e.g. dreaming, by day or by night). I’m guessing that at some point you’ve done something that you told yourself you wouldn’t. How could that have happened, if not by your brain sending your body signals that didn’t go through conscious processing? Which engine of motivation thus set you into motion? We could of course argue that you simply changed your mind a moment before acting. But, in my case at least, this doesn’t quite ring true; at times, the conscious desire to resist persists right through the offending acts. Passion, or lust, is a good example. There is a certain moment at which, empirically, things become inevitable. It can be felt throughout the body, a bit like the adrenaline rush of a fight-or-flight response, and it’s safe to say that it renders conscious thought irrelevant.

If we act without thoughts, and think without active control, is it possible that our actions are for the most part spurred by something other than cognition? That, like the elements of Aristotelian physics, we simply gravitate towards some desired state and subsequently create the narrative that portrays a conscious decision. I’ve often heard it said that fish don’t feel pain, let alone fear. Nevertheless, they clearly try to avoid undesirable situations (e.g. being caught). We call this “instinct”, but how do we know it’s different from what makes us run away? Perhaps the only unique trait of human intelligence is that we’ve somehow acquired a module for creating narrative, and thus our experience feels special. This would make human thought unique the way a child is; it has traits that make it unlike any other child in the world, but at the same time it’s just one member of the class “children”.

It’s easy to imagine a goldfish unable to understand our thoughts, but does that imply a clear-cut distinction where we have thoughts and it doesn’t? It’s harder for us to turn the argument around and imagine the perspective of a more advanced intelligence, and perhaps that’s just the point. A proverbial alien observer might see us as more intelligent than the goldfish, yet not consider either one capable of real thoughts. This conclusion might offend us, but how could we argue against it if we accept the goldfish as unthinking?


The catalyst for this post was Elie Maksoud, who by the way takes pretty pictures.

Unthinkable Thoughts

‘‘There are wavelengths that people cannot see, there are sounds that people cannot hear, and maybe computers have thoughts that people cannot think.’’

Richard Wesley Hamming

Something that is simultaneously trivial and fascinating (to me) is that the deeds of some other person my age will be very different, and sometimes far greater, than my own. Trivial because it’s so obviously true, for all of us. But fascinating because this person has had the same period of time and an almost identical human body to work with, so their accomplishments must have stemmed from differences in their environment and/or thoughts they had that I did not. No doubt the former plays a role, but if we confine our analysis to a college classmate I think we can establish a role for the latter as well: Sitting in the same auditoriums, coming from a similar background, this person somehow achieves a different understanding of the subject matter (and the world).

We apply the term “genius” to those who make important realizations that escaped everyone else, and try hard to explain what made these people special. Why didn’t the Theory of Relativity occur to everyone? While such explanations often emphasize a special combination of talent and opportunity, it also appears that simple birth defects and accidents can produce genius-level ability in the otherwise unremarkable. Based on this, one might propose that our brains normally have barriers which block many thoughts from appearing. But what is the system that determines which thoughts we have? And as a natural extension of this, how do we set ourselves up to have the widest/wisest range of thoughts?

Your thought-subset

As so often happens, Paul Graham has an interesting comment on this: he argues that we’re unable to think clearly about things that are part of our identity (e.g. religion, ancestry, preference for Apple products), and so to expand the range of topics you can productively think about you need to minimize your identity. Thought provoking (*cough*), and a seemingly perfect philosophy if you’re a Buddhist inventor. But it does seem more like a surgeon than a full-on savior: even if true, it only tells us how to remove certain specific blocks from our thinking.

Another proposition comes from an instruction that I wish someone had given my undergraduate self. We students were frustrated with having to cram huge curricula in some courses, and often vented about the folly of closed book exams. The important thing was being able to find information on demand, not memorizing tons of facts you might never need, right? Well, kind of… Now that I’ve spent some time thinking for a living, it’s clear that most of our progress comes from connecting dots. That is, coming up with solutions based on multiple pieces of information. Sure, you could easily look up those same pieces of information, but if they’re not already in your head when you encounter the problem you miss out on the solution. Based on this, a big limitation to what we’re able to think would simply be the quality and quantity of dots already in our heads; the more you already know about, the wider range of thoughts you can have.

To me this seems quite in line with empirical evidence, although it’s also obvious that other factors play a role. For instance, there’s the person who knows a lot of facts but somehow can’t venture into uncertain territory. To quote Hamming again:

If you read all the time what other people have done you will think the way they thought. If you want to think new thoughts that are different, then do what a lot of creative people do − get the problem reasonably clear and then refuse to look at any answers until you’ve thought the problem through carefully how you would do it, how you could slightly change the problem to be the correct one.

 

So your ability to think clearly plays into it, as does the amount of knowledge you have to draw on. It seems to me that there’s also a huge amount filtering that occurs before thoughts even enter your consciousness. That is, your brain actually processes a multitude of thoughts for every one that you’re aware of, but most of them are discarded almost immediately. Whatever governs this filtering process must have a profound effect on at least our subjective experience on thinking. I don’t know enough neuroscience to say anything really rigorous on this subject, but intuitively it seems possible that the filter is simply synaptic patterns formed by your past experiences. On a side note, maybe that’s where déjà vus come from: subconscious processing leaking slightly into the conscious domain, so that when the thought is presented to consciousness proper it seems to have (indeed has) happened before.

Such a filtering mechanism would certainly constitute a type of biological limit to what thoughts you are able to think. But one could easily imagine more profound limits based on the physiological wiring of our brains. It would naturally follow that different wiring would allow different thoughts. Returning to the original quote from Hamming, I’m sure most would agree that our deterministically programmed computers can’t think human thoughts. But perhaps they (or future versions of them) can think a different type of thoughts, which we in turn aren’t able to.

In other words, the Venn diagram might look like this:

Your thought-subset,  advanced

You might find it difficult to imagine the types of thoughts a computer would have; I certainly do. In order to come up with a decent answer we would need to examine what constitutes a thought, which I’ll leave for another post. But lest we let the barriers in our brains censor the very idea of inhuman thoughts, I’ll end with this reminder from Schopenhauer:

“Every man takes the limits of his own field of vision for the limits of the world.”

Last updated by at .