Tilted Forum Project Discussion Community - View Single Post - Extracting Images Directly from the Brain

roachboy · 12-12-2008, 05:19 AM

i don't think this experryment does what it claims--the key is the translation software.
the problem here is that the design is circular---subjects were shown a sequence of characters/images and brain blood-flow was monitored throughout--based on that, it was possible to assign weights within a topological space---which would mean that the weighting for the letters "neuron" were inferred, and then (it seems) the results of this inference were used to generate the results. so basically, you have an input (a letter) and a sequence of physiological changes associated with the shift into it (from the previous input) and out of it. so these changes are correlated with the input, some to operate inside the translation software as an analogy for the input in visual space (these changes then are another set of defining predicates, like the physical attributes which are rendered salilent through the category/name assigned it)...this correlation puts the model in a position to *produce* results---but the building procedure enables the claim that this is in fact a re-production.

but if you think about it, it is and isn't. so this isn't exactly peering into one's skull--this is constructing a model and then using that construct to model.

it's not obvious that representation operates in this way--it's not obvious that cognitive acts are entirely brain functions and not embodied---it's not obvious at all.

but this is interesting nonetheless...it reinforces a very old (and old-fashioned) notion of epistemology--which supposes that the world is simply given and knowledge of the world based on a duplication of that natural order. at the same time, i wouldn't worry about implications for privacy and the like simply because the results follow from the model rather than from what the brain actually does...

12-12-2008, 05:19 AM	#6 (permalink)
roachboy Super Moderator Location: essex ma	i don't think this experryment does what it claims--the key is the translation software. the problem here is that the design is circular---subjects were shown a sequence of characters/images and brain blood-flow was monitored throughout--based on that, it was possible to assign weights within a topological space---which would mean that the weighting for the letters "neuron" were inferred, and then (it seems) the results of this inference were used to generate the results. so basically, you have an input (a letter) and a sequence of physiological changes associated with the shift into it (from the previous input) and out of it. so these changes are correlated with the input, some to operate inside the translation software as an analogy for the input in visual space (these changes then are another set of defining predicates, like the physical attributes which are rendered salilent through the category/name assigned it)...this correlation puts the model in a position to produce results---but the building procedure enables the claim that this is in fact a re-production. but if you think about it, it is and isn't. so this isn't exactly peering into one's skull--this is constructing a model and then using that construct to model. it's not obvious that representation operates in this way--it's not obvious that cognitive acts are entirely brain functions and not embodied---it's not obvious at all. but this is interesting nonetheless...it reinforces a very old (and old-fashioned) notion of epistemology--which supposes that the world is simply given and knowledge of the world based on a duplication of that natural order. at the same time, i wouldn't worry about implications for privacy and the like simply because the results follow from the model rather than from what the brain actually does... __________________ a gramophone its corrugated trumpet silver handle spinning dog. such faithfulness it hear it make you sick. -kamau brathwaite Last edited by roachboy; 12-12-2008 at 05:23 AM..