You have to completely understand what something is doing internally to model it. Not just what happens externally as a result of internal activity.
Not really. Some auditory phenomena act like they were being processed through a Fourier Filter. We actually know enough about the auditory system and how it processes information to know that that isn't the case. Yet using such a filter in an auditory processing model mimics auditory system results quite well and it is a very economical model. So to with Laplace analyse for auditory signal envelope processing.
You just don't know enough about modelling or processes to understand the complex relations that exist between the two nor the way 'works like' tools can be applied.
As a student I actually modeled neural processing at the ear and produced results that looked just like those obtained by recording by only taking what was appearing as recorded output at each stage of transduction through to impulse production. 'Works like' is a very powerful way to model operator performance for instance. We can substitute capacitive storage and inductive filter equations for signature time constants for auditory processing from impulse through to continuous particular output.
Quite honestly if we only take output spike rates we can successfully model each stage of auditory output form every nucleus way station in the auditory system. Don;t even have to know how spikes are created nor sustained even though we have known that for about 70 years now.
If one looks across physical systems one can replace constants and times to use electrical current flow through a system of various wire diameters and produce water flow equations though pipes of differing sizes using the same basis equations. Even an elastic spring model can be used to model traffic variation a choke points in a freeway.
If they work the same way they usually can be modeled with the same eguations.
As far as I can tell the only situations where incomplete knowledge permits claiming something for nothing, like emergence, is inappropriate. Complete knowledge makes it obvious that if all interactions and attributes are known what is claimed emergent is reduced to determined.
Consciousness is system that can be treated "as like". It has served for subjective experience for millennia. We produce a system we know because it is what we experience. It is not as it really is which would probably take a more complete understanding of underlying factors. On the other hand we can model the neuron, neural networks, performance, prediction. approach and avoidance, et cetera, using models of what we know about the underlying structures and performance to build models we can test and verify whether what we know needs to be changed to produce better models or whether what we know is incomplete.
For instance we now know that the more or less uniform physical structure of cortex is used to take on tasks from other senses to produce working models of the world for individuals with loss of brain, tract, or receptor. We know people carry out localizing and object identifying when blind using tactile and auditory inputs. Now we know they do this by using visual cortex to accomplish these tasks using input from other senses.