Scientists Reconstruct Brains’ Visions Into Digital Video In Historic Experiment

Wow indeed! That is fascinating!!!

My goodness. I don't know what else to say. Well, othan than to say that this mind-blowingly cool.

Watching that dropped me into a state of don't-know-mind, big time. It's kind of unsettling, in a refreshing way.

Yeah, WOW.

So, this experiment claims (and I don't mean that in a pejorative way) to have recorded brain activity as the visual stimuli was presented to the visual cortex of each subject. This would show (hopefully) what images look like from a subjective perspective while experience is happening in the immediacy of the present moment. As this technology is refined in the future, I hope they will be able to compare a stimulus with the digital projection of the subjective representation to see just how accurate immediate perception really is.



Even more than THAT, I would like to see the same technology used to capture what one perceives subjectively during memory recall. Cognitive scientists have known for a long time that what we remember from an experience is largely constructed and influenced by past experiences. I image the further removed one becomes from the immediate experience of perceiving the stimulus, the less their subjective experience their memory of the event will match, say, what might be picked up from a camcorder.



Fascinating stuff!

Ooh, and one more thing!



It would be incredible to be able to recreate the subjective images of an individual who experiences delusions, such as audio or visual hallucinations.



There's also the potential to peer into the minds of people who report clairvoyant abilities. You know, to see if they really do "see dead people."



The potential applications just keep coming to me.

I would love to see how it recorded a vision, for example (imagery seen in a trance state). I hereby volunteer... Also would be neat to see how much more clear the imagery can be made when their database palette is larger.

Yes, a palette that is both larger AND higher resolution.

...
There's also the potential to peer into the minds of people who report clairvoyant abilities. You know, to see if they really do "see dead people."
...


-awouldbehipster

I "see" "dead people" but even if you saw what I saw, via a digital image machine, how is it possible to determine whether that's "really a dead person" or just subconscious imagery in my mind? Are dead people hanging out in some alternative universe, still dressed in their Sunday best from 1876? I kinda doubt it. I'm not sure what to believe about afterlife stuff or the ongoing coherence of a "person" after death in any way at all, but for some reason that doesn't mess up the interestingness of the experience.

You're right. "Seeing" the "dead people" wouldn't be enough. In order to support a hypothesis that someone actually communicates with "dead people" via their mind, more information would need to be gathered. Perhaps communicating with known deceased individuals in such a ways as to collect very specific data that only they would know. That, and it would have be something that could be tested over and over and over again, under sufficient controls.

But even without all of that, it would still be neat to see someone else's visions

You know what WOULD be really interesting to capture? Remote viewing in real time.

Indeed. I played a game with some friends last month where we took turns remote viewing. The video at the top of this thread reminded me of that, because often the images I saw were very impressionistic. I tended to see/think of qualities of objects, rather than specifics. For example, in one instance a person asked "what is under my bed" - I immediately thought of a large flat box and a large green snake. What he had under his bed was a guitar case and a coil of green wire. So the general qualities (flat, rectangular object, green coiled object) were evident but not specifically enough to say "guitar case" or "coil of wire". There may be people out there who are very skilled at this sort of thing and see very clear images. I'd never really tried it before, so I just tended to say whatever words or flickers of imagery came to mind.

What I want is for the thing to make a movie from visions and dreams. I suspect the raw imagery is not nearly as vivid as we interpret it to be - that is, there is a level of raw imagery that is quite rough and not very story-like, and then the narrative function or other functions back-form a storyline over it to string the raw material together. Have you ever noticed the dream content distinctly from the narrative content, such as when just falling asleep or waking up?

That headline is extremely misleading so I'm going to throw some cold water on the thing
As I watch the original video (four times now) and then the three subjects' "reconstructed" versions from a YouTube database of videos I was struck by how divergent the reconstructed videos were. Sure, some are pretty close. For example, it's pretty clear that the Berkeley software can recognize a human face -- but WHICH FACE? It also substitutes a woman's face for a parrot. How do we decide which is the right one? It would require the subject's opinion. May as well just ask the subject. And if for some reason you can't you are left with a probability that appears to be approximate, at best.
Yeah, it's pretty cool stuff but it's not what the headline asserts. It is not reconstructed video from mental images. It's pattern matching of the inference between blood flow in the brain and *potential* statistically probable images matched from a database of actual( but only potentially similar) video clips.
It certainly has some promise but I can't imagine how it could ever REALLY recreate the internal imaging of the brain as it relies entirely on existing video.
I'm wondering how the software would do if the subject were to IMAGINE a movie scene, not be watching one.
Sorry, I'm just sayin'


-cmarti

That's why they need a bigger database. I don't see why that's inaccurate, though, if the database were bigger? What are the chances that if you show 1000 people the image of a car, they see the mental image of a car? And if they are asked to imagine the image of a car without seeing one, they see the mental image of a car? I think statistical probability is enough for that.

I say imagine should be the same as watching, because they've done studies on amputees and such where if they imagine moving their (non-existent) right arm, the same parts of the brain light up as a person who is moving an existing right arm.

But sure, the headline is a bit tabloid. All headlines are. No one would read it if it were really long and accurate.

Two tips for a happy life:

Lower your expectations.
Fake it 'til you make it.

Works like a charm.

Excellent remarks, Chris.