4 ms·
One might wonder how Predictive Processing/Coding works when the "cortex" is empty (i.e., at birth). I don't know if this is just armchair neuroscience wrapped
by xpuente 4y ago
One might wonder how Predictive Processing/Coding works when the "cortex" is empty (i.e., at birth). I don't know if this is just armchair neuroscience wrapped in a bunch of mathematical flourishes.
- MarkusQ 4y agoThe cortex is never "empty". The blank slate model is an outdated speculation with no empirical support, akin to "you only use 10% of your brain" or "you can tell things about people's personality by feeling the bumps on their head".
- xpuente 4y agoI'm not sure that there is no empirical support: see [0] [0] A. Roe, S. Pallas, Y. Kwon, and M. Sur, “Visual projections routed to the auditory pathway in ferrets: receptive fields of visual neurons in primary auditory cortex,” J. Neurosci., vol. 12, no. 9, pp. 3651–3664, Sep. 1992. PS: I'm not saying that there is uniform initial connectivity. Just, the prediction level units (cortical columns) are unable to predict anything until they learn "something".
- MarkusQ 4y agoAs that paper's abstract says, "Cells in A1 of rewired animals differed from cells in normal V1: they exhibited larger receptive field sizes and poorer visual responsivity, and responded with longer latencies to electrical stimulation of their inputs" -- that is, the aditory cortex expects auditory data, and although it can adapt it doesn't handle it as effectively as the visual cortex does, because it's expectations aren't met. So that's evidence that there are predictions baked in from day one.