Intuition Machine
Published in

Intuition Machine

The Lazy Brain Hypothesis

Photo by Colin Watts on Unsplash

Amdahl’s Law in parallel computing says that you can only have sub-linear speedup when converting a sequential algorithm to a parallel one. One argument why the brain can do what it can is that it is massively parallel. But even though this may be true, this parallelism not imply that it performs fewer operations that conventional computers. There is a correspondence between the number of operations and power use. The more operations, the more power is consumed. But, why does the massively parallel biological brain do perform so many operations with so little power?

It is universally true that if you convert a sequential algorithm into a parallel algorithm, the number of operations increases for the same problem. So how does a parallel brain do more with much less?

The secret is that one way to reduce energy is to not do any work at all! A brain does what it does because it avoids doing any work if it can. Call this the “lazy brain hypothesis”.

Most of what the brain performs is based on ‘amortized inference’. In other words, behavioral shortcuts (or heuristics) that it has learned through experience and billions of years of evolutionary fine-tuning. A brain is an intuition machine.

But what is the algorithm such it really doesn’t do as little work at all? The lazy brain employs a lot of contextual information to make assumptions on what it’s actually perceiving. In fact, it’s isn’t even bother to perceive. It just hallucinates what it sees. The brain predicts what it perceives by running a mental model where it intermittently compares with the outside world. This is the inside-out theory of the brain.

The brain only fires up extra processing when it notices a discrepancy in its expectations. We call this a surprise (which incidentally is similar to being horrified). It’s a wakeup call for our consciousness to be engaged to do more work. Work required to resolve the surprise.

So when it is in surprise, it is executing more threads to quickly discover knowledge to compensate for the discrepancy. But the first one that gets a match implies that all other threads are shutting down. There’s no need to do additional work as long as a good enough response is reached.

In short, the parallel process does not do all the work that a comparable sequential process does. This is one explanation as to why the brain's parallel machinery doesn’t need to compute as much. The reason why this laziness is acceptable is that it only needs a good enough result and not the best result.

The massively parallel architecture of the brain allows for a massive number of threads to be executed, but most of these threads are terminated very early. But again, it is also a mistake to think that all the threads are activated.

Only the threads that are related to the current context are running. Furthermore, all threads are automatically inhibited within a fixed period. But the brain still maintains longer-ranged contexts because there are neurons that just happen to work at a very slow pace. So we can think of a brain having parallel processes running at different speeds, all providing context to one another. All basically not doing much for most of the time.

But what does ‘not doing much’ actually mean? It means that the brain operates on its default mode with as little energy as possible. Anything outside the normal requires energy.

What’s interesting is that our attention is controlled in the same way as motor actions (via the Basal Ganglia). Our attention navigates or feels its way just like your fingers feel its way while examining a fabric. The brain does the laziest thing with attention, by inhibiting all information leading to the cortex by performing the filtering at the thalamus. So our cortex never perceives information that it is attending to. It simply isn’t there to process. It has been filtered away by the Thalamus.

So when we are focused in thought, we are always on a very narrow path of perception. Attention seems to counter the emersion in the world with all one’s senses. To maximize our perception we dilute our focus. You can’t attend to everything, at best you attend to nothing! To engage wholely in this world, you avoid inhibiting your senses and that implies not attending to anything. It is in the same state as if you were in play.

Paul Simon explains the first words in his famous song ‘The Sound of Silennce’. The song begins like this, “Hello Darkness my old friend, I’ve come to talk to you again”. His explanation isn’t something profound. What he said was that he would go to the bathroom and turn-off all the lights to allow him a spaces without distractions so as to write songs. Our minds can be focused when we limit distraction (not the other way around as proposed in open offices).

However, when we are performing reason (i.e. system 2) our attention focuses on our thoughts. We are generating reasons after we perform our actions. We don’t reason and then act, that’s not the lazy way. We act, then we reason if we have to. Brains make up explanations after the fact.

The human eye has an actual blindspot, however when we don’t perceive this blindspot because our mind fills in the details. Our eyes have a fovea that can see in much greater detail as the rest of the eye. In fact, it is also why we can see color. If you put out your thumb in front of you, that is the fraction of space that the fovea can see. Yet, we see an entire field of continuous high-resolution space. How? Our brain makes up all the details outside of the fovea. The brain in fact does not spend time actually filling out all the details. Why should it? If there’s any detail that your attention might need, then your eye will move to focus its fovea on the detail. Why record something if you can just take a look at what it looks like in real life?

Cognition is all ‘Just-In-Time’. Why do anything when you can postpone doing it for later when you actually need to? This ‘Just-In-Time’ also is an explanation for Andy Clark’s Extended Mind hypothesis. Clark argues that the external affordances that we have to record our memories should not be distinguished as independent of our minds. Taking away the familiar environment for someone who has his memory impaired is like depriving that person access to part of his mind. We are in essence symbiotic with our environment and therefore it is important to be aware of how dependent a person is to external cognitive devices.

As we explore this lazy brain hypothesis, we begin to realize that our cognitive behavior works in a way that is the opposite of how we commonly think of it.



Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store