Quite. In fact, the data explicitly disproves any kind of grandmother cells. Consider for a moment the “Jennifer Anniston cells” that were reported in the media several years ago:
-
-
Replying to @tyrell_turing @neuroecology and
The researchers were recording from a *very* small subset of all neurons in the temporal lobes, and presenting ppl with a *very, very* small set of stimuli from the set of all possible stimuli. Yet, they found cells that responded to their stimuli. What does that tell us?
4 replies 0 retweets 12 likes -
Replying to @tyrell_turing @neuroecology and
We’ll, Rodrigo Quian Quiroga explained it very clearly. The Jennifer Aniston cell was actually a “Rachel in Friends” cell that also fire, at lower rate, for Monica and Phoebe.
3 replies 0 retweets 6 likes -
Replying to @apeyrache @tyrell_turing and
Exactly what you would expect in neural network with winner-take-all localist output units...
1 reply 0 retweets 0 likes -
Replying to @GaryMarcus @apeyrache and
No, I disagree. Given the sampling issues I discussed (small % of stimuli and cells) it’s actually a guarantee of a distributed code.
1 reply 0 retweets 9 likes -
Replying to @tyrell_turing @apeyrache and
but how would the data look different? given the sampling that we can currently do, there is no pattern of data that you would accept
1 reply 0 retweets 0 likes -
Replying to @GaryMarcus @apeyrache and
If we it was hard to find neurons in higher-order areas that responded to stimuli, and when they did they only responded to one (note: cells in these studies usually responded to multiple stimuli), that would at least not falsify a localist account. The current data does.
1 reply 0 retweets 1 like -
Replying to @tyrell_turing @apeyrache and
there are many connectionist models with localist output schemes (eg a node cat, anode dog etc) w variable activity levels that are thresholded by winner take all that behave exactly like this. localism is about what nodes stand for, not whether they have firing rates or real
#s1 reply 0 retweets 0 likes -
Replying to @GaryMarcus @apeyrache and
Those localist output schemes are kludgy, and NN modellers know that. A human’s output for ‘cat’ is very high D! And I never said it’s about firing rates: it’s the fact that cells respond to multiple distinct stimuli that disproves localism.
3 replies 0 retweets 1 like -
Replying to @tyrell_turing @GaryMarcus and
sorry I've lost track here. what are you trying to argue
@GaryMarcus? That there exist neurons in IT cortex that respond solely to one stimulus and no others (and maybe also that no other neurons in that area respond to that stimulus)?1 reply 0 retweets 1 like
they certainly are consistent w localist schemes that are standard in neural nets. more broadly: I wrote an article w Christof Koch: https://www.technologyreview.com/s/528131/cracking-the-brains-codes/ … plural in the title is key. i suspect brain uses many codes some local (perhaps groups of neurons); some distributed.
-
-
Replying to @GaryMarcus @tyrell_turing and
can you say *precisely* what you are claiming? This is where I'm confused
1 reply 0 retweets 2 likes -
Replying to @neuroecology @tyrell_turing and
not sure what you are asking. i am not saying that Aniston neurons never fire to anything but Aniston, but that they behave like localist WTA nodes. and that we can’t possibly rule out the existence of such things, based on very sparse data & poor understanding of neurosci
0 replies 0 retweets 0 likes
End of conversation
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.