Suppose a boulder rolls down a hill and smashes into a house. That boulder and the house are made up of molecules, and we could in principle fully describe the boulder smashing into the house in terms of the molecules the boulder and the house are made of, fully describe the interaction and how the molecules move and change, and never mention the boulder or the house.
So do the boulder and the house have no causal effects? That seems strange to say. Sure, we could describe the boulder as molecules, but changing the description isn’t eliminating ‘the boulder’. The boulder caused a hole in the wall of the house; the house caused the boulder to stop rolling.
Similarly, we can describe human behavior in terms of the chemicals and electricity that make up our brains. But if I want to understand why a kid is crying, it’s useful to use the cognitive shorthand of “she fell and scraped her knee and it hurts” than to describe the functioning of every one of the billions of neurons that result in that behavior.
Where it gets confusing is when we observe ourselves, and apply the same cognitive shorthand to our own thinking. Parts of the brain ‘observe’ each other, in the sense that they are wired up to receive information from each other.
When we think of a square, we have an experience similar to seeing a square because a part of our brain involved in seeing a square with our eyes is sending signals similar to what it sends when we see a square with our eyes. That part of our brain can be triggered by a pathway that begins with light hitting our retinas that was reflected from a square, but it can also be triggered by a pathway that begins with pressure waves that encode the words “picture a square” hitting our eardrums. The same part of our brain is triggered by both, and another part of our brains observes that with the cognitive shorthand of “visual experience of a square”.
And importantly, the cognitive shorthand has clear causal effects on our cognition. To use a ridiculous example, we understand what’s happening when Wile E. Coyote runs into a wall painted to look like a tunnel: the light hitting his retina triggered ‘tunnel’ rather than the ‘wall’, and so he acted as though he saw a tunnel. Less silly is the schizophrenic who hears his inner monologue and believes someone is beaming voices into his head.
That kind of mistake happens because parts of our brains rely on the signals sent by other parts of our brains, and that reliance affects the behavior of the part of the brain receiving the signals.
At the highest level, our cognitive shorthands about ourselves, our thoughts and feelings, our identity, our motives, all feed back into our brain to shape its connections and ultimately tune our behavior. That’s the observable causal effect of those descriptions.
And I think something similar applies to the computer without a monitor as well. A computer can process a JPEG and pattern match an elephant, and trigger a sorting function that puts it in the ‘pictures of elephants’ folder. It can do that by converting the numbers and letters into pixel intensities on a monitor, using a camera pointed at the monitor to convert light intensities back into numbers and letters and pattern matching on those numbers and letters to recognize an elephant. Or it can just pattern match on the original JPEG. Either way, when it passes ‘elephant’ to the sorting function, the file gets sorted into the ‘pictures of elephants’ folder.