It's interesting, thank you for pointing it out. However, I'm not convinced it's a huge step. People have already been able to do this kind of thing (better, actually) for what people are seeing; it's not such a surprise that they can do something analogous while you're asleep. It's not clear that we'll *ever* be able to "watch a dream like a movie" (although it's also not clear that we won't). fMRI probably has limited resolution, and we're still waiting for the technology to replace it, which will depend on much greater understanding of the brain.

All that said: pretty interesting line of exploration. I guess one interesting thing to explore long term would be to see if they can tell whether the brain activity starts in the semantically loaded area of the brain they were analyzing here, or if it is triggered from lower-level areas. That is -- they can tell, in this study, that somebody is dreaming of (say) "a book." But "book" is already a concept, a pre-existing high-level idea that the dreamer has from the world. Does the dream come from the part of the brain dealing with such high-level concepts, or is it an interpretation imposed on lower-level experiences generated by another part of the brain?

It's not completely clear to me whether this kind of study can ever in principle answer that question. Maybe. But it's an interesting step, as you suggest.

If it did ever become feasible to view people's dreams (or thoughts generally), there would be a ton of legal questions that I imagine would be opened up. Any opinions on those, abge?

I'm not familiar with what you mentioned regarding sight, but this seems rather different. Being able to recognize discrete objects that are purely constructs of thought seems like something pretty new.

I agree we are a long way off from dream movies and that this isn't an indication that it can ever happen, but it does give on pause.

Could dreams be used to prove premeditation in a crime? Will we have to submit a dream sample like we currently have to submit finger prints and sometimes DNA? Although fascinating from a scientific sense, it's wicked scary from a humanitarian sense.

Abge - I only scanned the article, but if I'm correct they need to monitor brain activity while the person is awake, first.

So, essentially, they have to calibrate the system to identify waking brain activity, and then when you are asleep they say, "Oh look....same electrical pattern...must be thinking about Draug in a tutu."

Interesting, I suppose, but I think it's far less intriguing that it sounds. We've known for a long time that you can monitor the electrical activity of the brain, and that you can stimulate memories, smells, etc by stimulating a specific portion of the brain -- in other words, this spot *here* is a hammer in your brain -- so I don't think it's such a stretch to monitor a sleeping person and see a signal and pretend we know what the person is dreaming about.

it's not like they can monitor a sleeping person and project a movie of their dream - not that at all; not even close.

Also, it can't be verified - so you can't ask, "Are you dreaming about 'x'?" and verify your prediction.

All that being said...I do like the idea, even if I'm less impressed than i wanted to be.

As for this being some sort of invasion of privacy - since (my understanding) this will only work if the have calibrated the system to my specific brain, they will only be able to invade your privacy if you first allow them to calibrate the system to you.

@krellin

Yes, of course it requires calibration. However, having a machine that recognizes you're consciously thinking of a chair is very different than a machine that can recognize a dream in which a chair is a main component. Dreams have significantly more noise than a conscious thought.

Privacy issues were clearly referring to a future in which this type of technology allowed someone to read other's dreams in a more efficient way.

Well, I'm not sure if I agree. I don't know enough to disdagree for sure, but it was the visual cortex they were looking at, so it's unclear to me that they were detecting higher conceptualization and not just visualization (and then performing recognition on the latter). Remember there were only about 20 concepts they were looking for.

Also, the article in Science Magazine points out that they weren't actually looking at REM dreams, which are (apprently) physiologically different from these falling-asleep hallucinations that they saw.

I don't think they did it while the people were awake, krellin. They calibrated it by waking them up when they were having dreams and asking them what they saw.

"However, having a machine that recognizes you're consciously thinking of a chair is very different than a machine that can recognize a dream in which a chair is a main component. "

It's not clear to me that this study successfully honed in on this "main component" issue. If a chair was anywhere in the dream the people were saying, presumably they would answer "yes" when asked if they saw a chair, even if the chair was just what the giant lizard was standing on to eat them.

@semck

That's a very good point. I'm not really sure how REM dreams differ from sleep-onset hallucinations.

http://www.wired.com/science/discoveries/news/2008/03/mri_vision

In this article, they showed people 1,000 images while monitoring the brain with a real-time MRI and recording the brain activity. Afterwards, they had very high predictability of identifying the picture that the person was looking at.

I believe this is essentially what the dream monitoring thing is -- requires prior understanding of brain electrical patterns before you can predict anything.

Why they will never be able to read your mind: ""It's quite tedious to measure every possible thought you might encounter, then measure the brain activity for that," said John-Dylan Haynes, a Max Planck Institute neuroscientist"

@semck -- Oh - OK....same difference. it's not reading your mind/dreams. Without specific personal calibration, it's useless. Also, I wonder if your patterns would change over time? Or, once you have the pattern for Draug in a tutu, will that 8always* be Draug in a tutu?

Things have changed a lot in 5 years. Also, I stand by my statement that consciously seeing an image will have less noise than a dream/hallucination, although I obviously have nothing to back that up.

" Also, I wonder if your patterns would change over time? "

No idea. That's certainly an interesting question.

Here's the one I was thinking about before, abge. This one impresses me more, and I think it may fully explain why these people could do what they did.

Be sure to watch the video.

http://newscenter.berkeley.edu/2011/09/22/brain-movies/

I guess I won't be concerned about invasion of privacy until:
1. I am required to wear around a device that constantly monitors my brain activity
2. I also wear a device that monitors my eye activity and records what I am looking at
3. etc....

If they can do anything to force me to provide them the calibration data that they need, then I have already lost all my freedoms, and the invasion of my dream privacy is the least of my concerns!

On the flip side...this *does* in fact 100% validate the premise of a story that I have written which is collecting dust in a file of mine...lol Soldier gets injured, injected with nano probes that repair his body....but which also allow the military to monitor everything he sees, hears, does, etc....essentially giving that full calibration data for his entire body...after which they take over his body while he is sleeping using neural stimulation with the nano network and turn him in to an unwilling assassin...

If I could have access to this type of technology and could get a full set of calibration data for my brain, I would then once again indulge in a favored college activity and drop a heavy dose of acid one last time....just to see...

Wow krellin, usually acid makes people more open minded, not less....

I am open minded. I open to the idea that liberalism is destroying the country. Maybe you should take some acid, then you, too, will see the light of truth. :)

First, here's the article for anyone looking:
http://www.sciencemag.org/content/early/2013/04/03/science.1234330.full

semck, "it's not such a surprise that they can do something analogous while you're asleep."
Even if this is nothing novel as far as findings, it is a new application, and finding that the new application works in practice is interesting. One never knows until one tries, yes? So sure, not surprising that it works, but there are still kinks (otherwise, their methodology would produce near 100% accuracy), and future researchers can look at removing those kinks as steps towards viewing dreams like movies (even if that is, ultimately, an impossibility). Lots of things that are impossible have been worked toward in science and the world is better for the myriad discoveries along the way. I wouldn't be so quick to dismiss this study either. Who knows what applications it may have outside the realm of dream interpretation/ translation?


I don't see it in the article (maybe I missed it?), but abge, your assertion that the dream-state has extra noise must be at least partially true since they opted for the NREM dreams rather than the REM dreams. It may be that they chose NREM dreams over REM dreams since it is easier to wake someone from NREM sleep ("We focused on [NREM dreams] ... because it allowed us to collect many observations by repeating awakenings..."). However, they also mention REM periods ("Reports at awakenings in sleep-onset and REM periods share general features..."). Anyway, here's what I found indicating there are differences between the two dreams:
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2629609/
http://www.sciencedirect.com/science/article/pii/S0361923004000930
and lastly, this article: bit.ly/10AquGF
Basically, NREM is boring stream of information, REM is memory/movie quality stuff.

"I don't think they did it while the people were awake"
Fairly certain they were awake.... "We constructed decoders by training linear support vector machines (SVM) (20) on fMRI data measured while each subject viewed web images for each base synset." Viewing material requires eyes to be open, which does not occur during NREM sleep.

Didn't follow the other links yet, but I think this topic deserves a bump, if nothing more than for the link to the full article.