Think about it like in the case of computers. The claim we are making is that our thoughts are essentially the same kind of things as the states that arise in a piece of software. And we already know that there exists today computer software which can be asked to modify itself through its own APIs.

Basically the way we view this is that CBT works kind of similarly to using the JVM APIs to modify some pieces of the running Java program to try to fix a bug. In this analogy, Psychiatric medicine would then be more like directly modifying the bits in RAM that represent a certain piece of executable code. They are both physical modifications ultimately, just working at different levels of abstraction.

They are not physical alterations. The software doesn't change the physical structure of the FPGA or CPU it's running on. It is not creating new gates or transistors or how they're physically connected, only utilizing the existing physical connections differently

Electromagnetic fluctuations in the wires are every bit as physical as the wires themselves.

There are always physical changes happening. External stimuli, and time based internal changes.

> If that's the case, the physical change should precede the changes in thought patterns.

Not precede, are. These aren't separate things.

> There's no chemical emission that can force someone to have an exact thought, dream, or inner dialogue.

How do you know that to be the case? It seems likely to me that an exact set of sufficiently complicated inputs and stimuli could generate an exact thought. Actually doing so would be too complex to figure out, but many physical systems are like that.

If it seems likely to you, then you're not as much of an empiricist as you claim