Psychological facts having physical correlates isn't the answer - to use a wrong-but-useful analogy; it's clear that hardware can inappropriately alter/damage software in many different ways, but in what ways software can/cannot damage or alter the normal functioning of the hardware? The other answers give some examples where this can happen - are there more direct ways?

I see where you're coming from, but in the case of biological brains and their physiology, the hardware IS the software, no?

Maybe not any that we are currently able to measure, but something's got to be changing. As an analogy, consider your computer: you could open it up, look inside, and feel that temperatures of different parts change over time, but you can't tell what it's doing or how well it is working just by looking at it. Even with fancy electrical equipment you'd have a hard time determining what's going on, beyond detecting that there's a lot of signal activity flowing around. To really see what's happening you need access through the software, and we just don't have that kind of access to the human mind.

It's not a question of being able to measure - I believe that we can measure e.g. synapses being strengthened and weakened as a result of learning skills; but that these changes don't change the physical functioning; it's just part of their normal functioning and it functioned exactly as well before and after - the behavior is different but physically equivalent.

For what I mean as a "meaningful change to the physical functioning", consider your computer - deleting all the software and installing a different operating system would not be a meaningful change to the physical functioning; the temperatures, electric charge and magnetic properties of the involved parts would be very different, but that doesn't matter as all the physical parts would still function in the same normal way although the screen would be displaying very different signals.

On the other hand, a software program that manages to physically disable or re-enable parts of some chip in a way that's not normally possible through software, blow a fuse in a hard drive or trigger a processor bug that "bricks" the processor - those would be the kind of changes to the general physical functioning I'm talking about. In what cases do we have such changes to the physical functioning of (parts of) the brain? Changes that are substantially different from any changes that the brain "does to itself" in the course of normal operation?

> Borderline personality disorder (BPD) is a common disorder associated with emotional dysregulation and other symptoms that have been hypothesized to be related to dysfunction of limbic brain areas including hippocampus and amygdala. The purpose ofthis study was to measure hippocampal and amygdala volumes in BPD. Hippocampal and amygdala volumes were measured with magnetic resonance imaging (MRI) in 10 patients with BPD and 23 control subjects. Patients with BPD had a 21.9% smaller mean amygdala volume and a 13.1% smaller hippocampal volume, compared to controls. These findings are consistent with the hypothesis that alterations in the hippocampus and amygdala are associated with BPD.

Here they see changes in the structure of the brain.

They see changes in the structure of the brain and they see changes in behavior (BPD), but they're clearly not saying that BPD caused these structural changes - they list both causal directions as some of possible hypothesis (altered amygdala/hippocampus causing BPD and/or BPD-related experiences causing these alterations in amygdala/hippocampus), in addition to other possible scenarios e.g. the fact that all of their BPD-group was on psychotropic drugs could also have been a cause for the observed changes, etc. They also aren't making any conclusions about a possible link between childhood abuse and BPD (or link between abuse and these brain changes) beyond the fact that they chose a set of 10 subjects that all had both BPD and suffered abuse; they're not even discussing a correlation as their data isn't sufficient to say anything about that.

All the observations in this paper certainly are compatible with a hypothetical scenario where these brain changes cause BPD (or indeed, are BPD) but it was some factor other than emotional trauma that caused the observed changes in their brains. I don't know if it is that way, but this particular paper also doesn't give any evidence to support or deny it.

[edit] It even seems that they don't even state that there was a change in their brains - instead, they observe a difference between BPD patients and the control group; but it would also be consistent with a hypothesis that these patients had smaller hippocampus/amygdala e.g. since birth, and simply having that property (instead of a change) was a risk factor predisposing them to suffering BPD symptoms later in life; and the authors acknowledge this as a possible option in their conclusions.

The question at hand is whether you can have thought patterns that cause structural/functional changes to happen.

On that note, memory's footprint is all over the brain. Learning judo is more than just filling up a psudo HDD with relevant information.

Granted this link fades over time (and likely therapy helps), but the body is not always good at handling major stresses and the damage can become permanent.