Often when I read papers on neuroscience, I find it difficult to dismiss an analogy. It feels much like we're trying to analyze how different models of computers work by giving them all some input and then trying to discern meaning from the circuitry that then activates in response. The problem with this is imagine I give you even the precise specs of a fairly basic computing system. You're going to be able to create a lot of correlations, yet you'd probably make effectively 0 meaningful progress towards 'cracking' the system, or really gaining any meaningful degree of insight beyond repeating correlations. E.g. it may be that if you press the 'f' key, a certain area of your circuit board sees a heat spike but that doesn't really tell you much of anything. And, at worst, can give you false leads as you start to draw correlations such as 'ahh!!! it heats up when I press f, g, and h, but not i, j, k!!' When the actual reason, as is easy to imagine, might be entirely spurious.

And in this case the analogy is many orders of magnitude worse. The brain is, by far, the most complex computing system we know of. And instead of precise specs, you have nothing but previous correlations to try to even have a clue as to what you're studying. And even of the specs we can measure, it's not looking hot. The Openworm [1] group for instance has been trying to model a worm brain. The roundworm brain is about as simple as you can get: 302 neurons, 7,000 synapses. The human brain's at 86 billion neurons, 100 trillion synapses. Yet even that worm project seems to have hit some unforeseen hurdles since it appears to have stalled out since making headlines some half a decade ago.

Of course neuroscience is far from my specialty, and it's entirely possible I'm missing some critical nuance. I'd love to know why this analogy is inappropriate if anybody could share.

[1] - http://openworm.org/