And you lose the ability to talk about probability in a meaningful way.

If something does create an almost unprecedentedly large response in the brain, does that make it more likely to create lasting effects in the brain and cause the formation of new neural pathways?

I don't know. You want to claim that it's highly probable but you've provided no evidence for this. I know that porn addicts brains are different, primarily in terms of the reward system in ways similar to other addictions I believe, but you'd expect that to be the case. And you've certainly not linked any probability of large responses to the probability of potential changes to the probability of any potential to harm.

If you can't put numbers on it to compare it to other things, you really have no way to talk about probability that's not just waving your hands around and going 'Look at the brain scan!' Well, so what? Even assuming that all of your premises are true, (and the probability of concurrent events is the sum of their multiples so you've lost probability at every stage whatever the actual numbers would turn out to be; complexity penalty) and these brain scans actually exist, what does that mean? Is the probability of harm five times as likely as something else, twenty? If the initial chance of harm is quite small this doesn't make much difference. Is it three orders of magnitude out?

Neurology is a complicated area of inquiry, unless you know how often a given change links up with some result, it's just a picture.

You can't forget about psychology and stats and still meaningfully make statements like 'it's not even on the same chart' you have to know where both things are to be able to talk about that sort of thing - you have to have put the work in and have the numbers.