A Trick of the Mind

 

A trick of the mind, by Daniel Yon, is a book about how the mind works.  The basic idea is that we don’t see the world “as it is”, but a combination of the inputs and the mental models we hold of it.

 

Mental models and neural networks 

Many kinds of AI have “neural networks” as their underlying technology.  The idea is borrowed from nature, where people figured out that there isn’t a neuron for (say) “red”, or any other particular concept, but rather that all the thoughts of “red” or anything else are embodied in the configuration of how the body is connected, and in particular, how the nervous system is connected.

Computer neural networks are trained by adjusting the connections between nodes.  This is analogous to the physical connections between neurons with each other (and with other tissues).  The connections form an implicit model of the world.

At the most basic level, the connections can determine what can be perceived at all.  There was an experiment that was done in 1970, in which kittens were allowed to see only vertical or horizontal lines in their early development.   When they were put into a normal environment the vertical only kittens would walk into table legs and the horizontal only ones couldn’t see vertical objects.  

This was important for medicine in humans and for understaning what is called neural plasticity, that is, the idea that the nervous system can rewire itself in response to the environment.

While the kitten experiment is looking at low level neural processing, Yon’s book describes the effects of the implicit models that people use to navigate their environments.

Models are necessary 

The first and most important point is that models aren’t just an accidental and possibly unfortunate aspect of thinking.  They are necessary to navigate the world at all.  Our brains are stuck in the head, with only the information from our senses to help them understand the world outside itself.  This information is sparse and ambiguous.  To take an example, anything you see is a 2d representation of a 3d object.   There are many ways a 3d object can be squashed down to 2d.  One question is just how far away it is.  Something that’s bigger, but farther away can look exactly like something closer but smaller.

The way the brain handles this problem is by having an idea of what an object is.  You have an idea, for example, of how big a truck is.  However, if you are at the top of a very tall dam and look down at the trucks at the bottom, while your intellect knows that they are full sized trucks, your feeling is that they are toys.  This is one of my memories of the trip my family took to get from Arizona to Seattle, years ago.

Looking down the Hoover Dam, Richard Gangemi, CC BY-SA 4.0 via Wikimedia Commons

 

Show me the evidence 

It makes sense theoretically that the brain would need to be seeded with clues about what it’s seeing, but does that actually happen?  That is what Yon studies, and spoiler alert, yep, totally.   

To give an example from Yon’s book, he talks about making sense of speech.  In written language, we have spaces and punctuation, as well as spelling, to help us understand what the words are.  In spoken language, he points out, words tend to run into each other.  

I’ve read comments where people have talked about some language as having just a few big words that are made up of smaller units.  I wondered then how they (the observers) could tell this, because isn’t that the way it works in, say, English?  It turns out that the minds of English speakers insert the word boundaries because we know where they are.  This happens unconsciously, so we aren’t aware of it.  In other languages (I think the ones I read of were perhaps ones without a written form), since the observers didn’t know the word boundaries, so it all seemed to run together for them.

The evidence, then, is that raw perception isn’t actually raw, but it very much pre-processed before it gets to consciouness.  But wait, there’s more … but in another post.

 

Note on the kitten experiment 

There isn’t a single online description of the kitten experiment, but here is the summary from the search.  Don’t yell at me about being mean to kittens, I’ll take info where I get it, even if kittens didn’t perhaps live their best lives ever. 

… Search summary ….

It is a landmark experiment in neuroscience conducted in 1970 by Colin Blakemore and Graham Cooper at Cambridge University.

The experiment provided early proof of neuroplasticity—specifically showing that the brain’s visual cortex isn’t fully hardwired at birth, but rather physically wires itself based on the environmental input it receives during a critical developmental period.

The Setup

• Darkness & Isolation: Newborn kittens were kept in total darkness for their first two weeks of life.

• The Cylinders: Starting at two weeks old, they spent several hours a day inside tall, smooth cylinders lined with high-contrast black-and-white stripes. One group saw only vertical lines, while the other saw only horizontal lines.

• No Other Visual Cues: The kittens wore wide black collars so they couldn’t even see their own paws or bodies, and there were no corners or edges in the cylinder.

• The rest of their time was spent back in total darkness.

What Happened When They Grew Up?

At around 5 months old, the kittens were brought into a normal, well-lit environment. The researchers observed striking behavioral and neurological outcomes:

Behavioral “Pattern Blindness”

• Vertical-raised kittens: They navigated normal spaces by walking into chair legs or table legs as if they weren’t there. However, if a rod was held vertically and shaken, they would play with it.

• Horizontal-raised kittens: They ignored vertical obstacles, but if a rod was laid horizontally, they would track it and pounce.

• Neither group reacted to objects or visual edges oriented in the direction they had been deprived of growing up.

Brain-Level Changes

When the researchers recorded electrical activity in the neurons of the kittens’ visual visual cortex (specifically the primary visual cortex, where “line orientation” neurons live):

• In a normal cat, neurons respond to lines at all angles (vertical, horizontal, diagonal).

• In the vertically-raised kittens, practically zero neurons fired in response to horizontal lines.

• In the horizontally-raised kittens, neurons only responded to horizontal lines.

The brain had physically pruned or repurposed the neurons that would normally process the missing orientation because they were never stimulated during the critical window of early development.

The Aftermath

While the kittens eventually recovered basic spatial orientation, depth tracking, and movement after living in a normal environment for a while, their specialized orientation blindness remained largely permanent.

Why it matters: This experiment (along with related Nobel Prize-winning work by David Hubel and Torsten Wiesel) revolutionized how we understand human vision. It directly influenced medical treatments for human infants—explaining why conditions like childhood astigmatism, congenital cataracts, or crossed eyes (strabismus) must be corrected very early in life to prevent permanent brain-level vision deficits.