Why Do We Dream?

Medium

Musings on memories and Darwin machines

In a recent post, I looked at consciousness from the perspective of the prefrontal synthesis model. In this view, our experience of the world is mediated by neuronal ensembles, collections of neurons that fire in synchrony when activated. ObjectNEs, as these circuits are called, organize our sensations according to the principle that ‘what fires together wires together.’ By their very nature, ObjectNEs consolidate the buzzing booming confusion around us, condensing our model of the world into a manageable number of perceptual building blocks.

ObjectNEs are active during sensory integration, but they can also be ‘pinged’ deliberately by the prefrontal cortex, the planning center at the front of the brain. The purposeful activation of a sequence of ObjectNEs is what we mean when we talk about a “train of thought”. ObjectNEs can also light up spontaneously in unexpected combinations, a process that we experience as dreaming or hallucinating. The signal to permit (disinhibit) spontaneous activation has been identified by sleep researchers as stemming from a posterior region of the brain at the border of the lateral parietal and occipital lobes, the so-called dream hot zone.

This gives us three scenarios by which we become conscious of ObjectNEs:

1) During the synthesis of new percepts from multichannel sensory input

2) During prefrontal synthesis, in which our command center activates a sequence of ObjectNEs as part of problem-solving behavior

3) During the spontaneous activation of ObjectNEs when they become disinhibited by the dream hot zone (primarily in REM sleep).

— —

The formation of ObjectNEs is directional, by which I mean that the effort expended is justified by the benefits to the organism. Specifically, these circuits compress data and make new building blocks available for modeling the outside world.

Likewise, the purposeful activation of sequences of ObjectsNEs by the prefrontal cortex is directional. The energy required is well spent because the resulting train of thought assists us in problem solving.

The bizarre nature of dreams and hallucinations, however, seems at first glance to be something else entirely. If dreams reflect the spontaneous activation of multiple ObjectNEs in essentially random combinations, the question becomes, “Why does the brain bother?” Surely there are better uses for our energy reserves, particularly while recovering from the stresses of the day during sleep.

— -

It occurs to me that the answer to this riddle might be found by looking at directional processes in nature that incorporate randomness. And the candidate at the top of the list is Darwinian evolution. Evolution is essentially a three-part set of instructions for increasing the order (decreasing entropy) in an open system.

Step 1 is to generate random variations of some seed phenomenon.

Step 2 is to apply a filter (natural selection) to these variations to eliminate all but the fittest results (the definition of ‘fittest’ depending on the context in which the Darwin machine is operating). And

Step 3 is to repeat the first two steps, using the filtered results as the seeds for the next virtuous cycle.

Looking at dreams from the perspective of Darwin machines provides multiple explanatory benefits.

1) Why do we dream? We dream because the brain periodically engages in a perceptual ordering process that can be described as a Darwin machine. When ObjectNEs are activated, we have a conscious experience, and if we are asleep at the time, we refer to this afterward as dreaming. Essentially, we are along for the ride.

2) Why are dreams anarchic? Step 1 in this recursive cycle is to generate random variations of the connections between our stored percepts.

3) Why are we biased to dream about our recent past or our preoccupations? The generation of random connections need not be entirely random. ObjectNEs could be weighted by temporal proximity or emotional valence, making them more or less likely to be activated.

4) What is the precise criterion for ‘fittest’ applied as the filter in Step 2 of this cycle? I don’t know.

Perhaps a poet or a neurologist reading this could suggest an answer.

If I were to guess, I would say that just as the body flushes out the toxins from the neuronal gaps during sleep, the Darwinian dream machine probably does something similar but on an informational level. A brain defrag if you will…

Toward a testable hypothesis: If dreams reflect the operation of a Darwin machine, it should be possible to investigate the process as follows: Record the first set of dreams for the night. Record the second set of dreams, and compare. The percepts in the second set should represent variations on a subset that appeared in the first set.

In addition to providing support for the Darwin machine hypothesis, this approach could be used to infer the nature of the criterion for ‘fittest’ previously mentioned.