Deep language algorithms predict semantic comprehension from brain activity

https://www.nature.com/articles/s41598-022-20460-9

by piecerough • 4 years ago

95 46 4 years ago

piecerough• 4 years ago

Here, we show that the representations of GPT-2 not only map onto the brain responses to spoken stories, but they also predict the extent to which subjects understand the corresponding narratives. To this end, we analyze 101 subjects recorded with functional Magnetic Resonance Imaging while listening to 70 min of short stories. We then fit a linear mapping model to predict brain activity from GPT-2’s activations. Finally, we show that this mapping reliably correlates (R=0.50,p<10−15) with subjects’ comprehension scores as assessed for each story.

notfed2• 4 years ago

You should use quotation marks when quoting.

tough2• 4 years ago

Most hella HN comment of the day

_s_a_m_• 4 years ago

This is so bizarr, I wouldn't even know how to inspect, verify and critique the claims made. Also there is no subject in such simple algorithms, very relaxed use of big words.

9wzYQbTYsAIc5• 4 years ago

Neuropsychology is what you may need to look into to fill in the gaps.

edit: https://mitpress.mit.edu/9780262680530/parallel-distributed-...

Nokinside3• 4 years ago

Nature Scientific Reports has very light peer-review. If you pay the fee, they publish with high probability.

Manuscripts are not assessed based on their perceived importance, significance or impact https://www.nature.com/srep/guide-to-referees#criteria

sxg• 4 years ago

I've published two papers in Scientific Reports and have had several more rejected—definitely not my experience. Peer review was arduous and on par with any other major journal with similar impact factor.

Nokinside3• 4 years ago

The acceptance rate for Scientific Reports is 55%.

civilized3• 4 years ago

Not accurate. There are journals that try to evaluate impact and there are journals that focus purely on scientific validity. (PLOS One is a journal that has become famous for doing the latter.) The passage you quote is just the journal signaling that it's in the second camp. It doesn't mean the peer review is any lighter than at impact-focused journals.

It's a good thing for science that not all journals are impact chasers. Scientists are by definition not perfectly reliable evaluators of impact, because science is about exploring the unknown. Publishing work that's only passed a technically focused peer review allows for unexpected impact.

Nokinside3• 4 years ago

The acceptance rate is >50% and it lets lots of fake and pseudoscience trough. https://www.science.org/content/blog-post/more-scientific-re....

It's a good thing for science

Maybe, but we should treat them like arXiv.org type e-Print archive. People are posting them to HN and thinking that because it's Nature.com site it's solid science.

theGnuMe2• 4 years ago

Pretty wild finding.

If it holds up, you could monitor kids in class, dementia patients.. wild. Start your startup engines.

Extrapolating, this also suggests neurolink should work and you can probably do it with less invasive tech.

9wzYQbTYsAIc5• 4 years ago

Don’t get too far ahead of the results, but yeah, this makes things extremely interesting.

From the paper:

”However, whether these models encode, retrieve and pay attention to information that specifically relates to behavior in general, and to comprehension in particular remains controversial”

edit: Mapping from a statistical model directly to one specific individual’s brain may turn out to be intractable for things like brain implants.

But, the chances are pretty good that there will be strides made in unexpected areas.

fnordpiglet• 4 years ago

As a manager I could hook it up to my corporate performance system and use ability to comprehend boring meeting content as a performance metric. It’ll be awesome, you’ll see.

samat• 4 years ago

There will be a hot startup for this and then MS will roll out buggy half-assed copy and will capture the whole market.

m463• 4 years ago

I can't help think -- will this be used to measure engagement for advertising, or the effectiveness of political speech?

tough2• 4 years ago

Yes, utopia or panoptcion

theGnuMe2• 4 years ago

Yep.

9wzYQbTYsAIc5• 4 years ago

Ah, this is truly beautiful - neuropsychology and artificial neural networks making connections.

From the paper: ”These advances raise a major question: do these algorithms process language like the human brain? Recent studies suggest that they partially do: the hidden representations of various deep neural networks have shown to linearly predict single-sample fMRI, MEG, and intracranial responses to spoken and written texts.”

9wzYQbTYsAIc5• 4 years ago

From the paper: ”Specifically, we show that GPT-2’s mapping correlates with comprehension up to R=0.50. This result is both promising and limited: on the one hand, we reveal that the similarity between deep nets and the brain non-trivially relates to a high-level cognitive process. On the other hand, half of the comprehension variability remains unexplained by this algorithm.”

c3534l• 4 years ago

Maybe I'm missing something, but I don't see how this correlation shows that the mapping is semantic and not, say, grammatical, syntactic, or structural in some way.

version_five• 4 years ago

It would be interesting to try the same experiment with a syntax parsing / part of speech tagging model and see if it holds.

9wzYQbTYsAIc5• 4 years ago

There is deep discussion of that topic in the paper. Attention, audio processing, and lexical layers are brought to the forefront of their discussion.

To quote the authors: ”We propose that deep neural networks encode a variety of features…”

mjburgess6• 4 years ago

The correlation doesn't even show a link.

Run GPT on other inputs, run fMRIs on other inputs: call that dataset, (G, F).

Now consider all possible subsets, (g_i, f_i) in (G, F) ...

How many show correlation in their choice of NN property, and their choice of brain property (ie., blood flow)?

My guess: *many*.

This is trivial to refute if you have any sense of the scientific method. Construct the converse hypothesis and test it. They didnt.

jcims2• 4 years ago

How long before a computer can see kernels of thought before we perceive them in our mind?

filoeleven• 4 years ago

Reminds me of this Ted Chiang short story.

https://www.nature.com/articles/436150a

jcims2• 4 years ago

Exactly! Seems very plausible.

civilized3• 4 years ago

101 subjects seems like not that much data to establish a correlation between GPT activations and brain activations, a correlation between brain activations and reading comprehension, and then chain them together to get an overall correlation.

glup• 4 years ago

I'm missing how this is an advance over https://www.pnas.org/doi/10.1073/pnas.2105646118

arpa• 4 years ago

Ooh, more mind reading machines in the age of total privacy loss, nice.

danuker• 4 years ago

For a u-/dys-topia where everything is public, check out Manna by Marshall Brain.

ggm• 4 years ago

Name checks out. His, not yours!

dlkf• 4 years ago

Brain scans, small sample size, sensational claims; when will we learn?

mjburgess6• 4 years ago

My first response on reading the headline, "no, they don't".

What's the point of Nature?

This article is the pinnacle of modern pseudoscience.

phonebucket• 4 years ago

Minor nitpick, but this is Scientific Reports, not Nature.

It's run by the same publishers as the journal Nature, but is a significantly lower impact journal.

kleiba• 4 years ago

Should be repeated with other LLMs for confirmation.

catelm• 4 years ago

First, we evaluate, for each voxel, subject and narrative independently, whether the fMRI responses can be predicted from a linear combination of GPT-2’s activations (Fig. 1A). We summarize the precision of this mapping with a brain score M: i.e. the correlation between the true fMRI responses and the fMRI responses linearly predicted, with cross-validation, from GPT-2’s responses to the same narratives (cf. Methods).

Was this cross checked against arbitrary Inputs to GPT-2? I gather, with 1.5 Billion parameters, you can find a representative linear combination for everything.

The Bible Code comes to mind (https://en.wikipedia.org/wiki/Bible_code).

mjburgess6• 4 years ago

They assume linearity. They map their choice of GPT-properties to their choice of (brain blood flow!) properties. They then claim there are correlations, with a few fMRI datasets.

If something serious was on the line, with this type of analysis, you'd be fired.

Reading this it feels like we might as well give up on there being any science any more, tbh. For this to appear in Nature -- it feels like the rubicon has been crossed.

How can we expect the public not to be "anti-vax" (etc.), or otherwise scientifically competent in the basic tennets of modern science (experiment, refutation, peer review) -- if Nature isnt?

feet2• 4 years ago

That's a pretty extreme statement about the entire field of scientific research being based off one paper

svara• 4 years ago

It's not Nature, it's Scientific Reports. The bar to publication in the two couldn't be more different. Nature is one of the premier high impact journals, Sci. Rep. is a pretty middle of the road somewhat new open access journal.

DEDBEEFF• 4 years ago

To this end, we analyze 101 subjects recorded with functional Magnetic Resonance Imaging while listening to 70 min of short stories. We then fit a linear mapping model to predict brain activity from GPT-2’s activations. Finally, we show that this mapping reliably correlates (R=0.50,p<10−15) with subjects’ comprehension scores as assessed for each story.

Note that this is exactly the wrong way to form and attempt to refute a scientific hypothesis. The authors don't start with some new observations that require explanation, they start with a hypothesis already fully-formed ("...these models encode information that relates to human comprehension..."), and then go out and collect observations to confirm this hypothesis.

I'm sure that if asked, the authors would say that they are simply trying to answer a scientific question, but it's obvious that they already have the answer they want and they're just trying to find data to support it. The problem of course is that if one is already convinced of the answer, one can always find evidence to "prove" it. It's a kind of confirmation bias.

Deep language algorithms predict semantic comprehension from brain activity

Related Stories