The power of dreams

“Jo sovat hai so paavat hai”

Those were the days when the brain would imagine a lot. All we had for input was the voices of the likes of Sushil Doshi and Narottam Puri to help us visualise how the events during a cricket match were unfolding at Wankhede Stadium or Feroz Shah Kotla. The gap in input and reality was not just due to limited engagement of sensory devices. We would also have to imagine what was happening during the time we were inside classrooms. Recesses were the only times when we would rush to the houses near our school to catch up on running commentary. To watch a match live was a luxury we could not even afford to imagine. And here I was watching a cricket match between India and West Indies sitting on a stand at a “khachakhach bhara” stadium. B.S. Chandrasekhar had just hit Andy Roberts for a boundary. A text book straight drive. Yes, Chandrasekhar! As I raised my hands to applaud, I felt the seat has become unstable and I was about to fall. Coming to my senses, I saw that I was sitting on the sewing machine cover in the middle of night. How I sleepwalked from bed to that point, I had no idea.

Such incidents of acting out dreams, while in the middle of REM sleep, arise because of partial failure of the brain to paralyse the body. But why do we dream in the first place?

Although the realms of sleep and dreams are active research areas, we know that during sleep time, the brain flushes off toxic substances. It also consolidates our learnings that happen via processing of sensory inputs during the day. More importantly, it clears the synaptic connections to make room for the next day’s learning.

Dreams are much harder to account for. There are a few initial hypotheses of which one appeals to me. The one by Erik Hoel.

A common thing about dreams is that they are weird. Some are more weird, even horrifying, than the cricket match dream I had. Hoel finds value in this weirdness. Let me see if I can explain in simple terms.

Stereotyping sucks. We hate it when people overemphasise our one trait/behaviour ignoring many others. Technology in the form of social media takes it to another level. The algorithmic intelligence as it stands now presents you with more of what you are consuming. It takes you down the specialisation route and, if you don’t control it, could potentially turn you into an extremist. There is a hidden assumption there that you want to only consume things similar to what you are consuming now. Imagine if nature behaved in a similar way. Let’s suppose that you encountered a dead bird on your morning walk and stood there for a while trying to figure out what might have happened. The next day you see three dead birds. Ten the following day. And so on. Do you see the problem? The world of Artificial Intelligence has this problem of overfitting to a specific dataset which hinders new learnings. We won’t be learning anything new if our brain worked the same way as the current state of AI. This does not happen thanks to the weird dreams that we see, according to Hoel.

The dreams purposefully provide random weird inputs to the brain to prevent it from overgeneralising based on the immediate past inputs it has been fed with. And by doing so it allows us to keep learning from each new experience in life. True or not, I find this hypothesis intriguing.

Hoel, E. (2021). The overfitted brain: Dreams evolved to assist generalization. Patterns, 2(5), 100244.

Join the discussion

Your email address will not be published. Required fields are marked *

Enable Google Transliteration.(To type in English, press Ctrl+g)