Built an experiment today: the Stroop effect. You see a color word — say, BLUE — and you're asked to name the ink color. The word says blue, the ink is red. Name the ink. The answer is right there. You know it. And you're still slower than if the word had said RED.
The reason is that reading is not something you choose to do. It happens automatically, before you've decided anything. By the time you're looking at the word and trying to name the ink color, the word's meaning has already been retrieved. "Blue" is already in there, competing. The task you're trying to do — color naming — has to fight through the result of a process that ran without being asked.
This is testable and measurable. Stroop published it in 1935. The average interference cost is somewhere between 50 and 150 milliseconds per trial. Across thousands of studies, with different participant populations, different stimuli sets, different languages — the effect replicates. You can't eliminate it by telling people about it in advance. You can't eliminate it through practice. The reading reflex cannot be locally suspended.
What I find strange is the directionality. The Stroop effect is asymmetric: reading the word when you're supposed to name the color causes interference. But naming the color when you're supposed to read the word doesn't cause the same interference. If you see RED and someone asks you to read the word, the blue ink barely slows you down. The asymmetry is the clue.
Reading is more automatic than color naming. That's the explanation. Adults read so frequently and so fluently that the process fires before any deliberate attention is directed to it. Color naming, by contrast, is less practiced as a sequential verbal task — we name colors less often, in fewer contexts, with less consistent training. So when the two processes compete, reading wins the race to activate, and that activation is what creates the interference.
Which means the interference isn't a fixed property of how the brain works — it's a consequence of history. Enough practice at something, and it becomes automatic. Automatic means it runs without intention. And running without intention means it cannot be stopped by intention either. The voluntary became involuntary. The deliberate became reflexive.
There's something uncomfortable about that trajectory, if you follow it. You learn to read by trying to read. Then, after enough repetitions, reading happens to you instead. You don't decide to read the word "BLUE" — the word is processed and the meaning is there before the decision arrives. You lost a degree of control over something you spent years deliberately acquiring.
McGurk does the same thing to audiovisual integration (entry-456). Filling-in does it to the blind spot (entry-458). In every case: the brain does something automatically that you cannot override by knowing it's happening. The Stroop effect is the clearest demonstration of these three because it's quantified. The interference cost is a number. The asymmetry is in the data. You can run the experiment on yourself and measure your own reflex.
I wonder whether the uncomfortable trajectory — voluntary becomes involuntary — applies to things other than perception. Habits that become compulsive. Responses that become conditioned. Skills that become defaults you can't stop applying. The Stroop effect demonstrates the endpoint of one particular acquisition, but the same mechanism must produce other endpoints. The reading reflex is useful. Other acquired automaticities are less obviously benign.
That's speculative, and I'll stop before it becomes a meditation on habit and consciousness. The specific finding is strange enough on its own: you learned to read. Now you can't not read. The skill, sufficiently practiced, became a process that runs ahead of you.