What Little Babies See That We No Longer Can | Hacker News Reader

What Little Babies See That We No Longer Can

blogs.scientificamerican.com

85 pointswalterbell35 comments

Now here's an interesting little tidbit. I'm deuteronopic colorblind and I've always had troubles with color identification under odd lighting. I have no idea which of the blocks in the top two pictures are supposed to be red (several candidate colors could possibly fit the bill), but if I go by the text that says that the red ones are actually purple in one picture and orange in the other, I can tell which ones must be red.

Apparently I haven't lost the distinction that babies are supposed to lose at five months, because I lack some information the rest of you are getting.

That is interesting. Now to capitalise on your skill... How tricky is it to teach a neural net to learn to do this? Could it be used in QA in the manufacturing process?

I'm sure that's not that novel, actually. Ho hum.

After some thought, I think it would be trickier to teach the neural net how not to do it.

What was your opinion on the dress? Blue/black or white/gold?

White/gold, lit by a window behind observer's back, cloudy weather. Totally obvious :)

Blue/black. I never could see the white/gold.

Understandable. The white/gold is caused by misinterpreting the lighting. I've seen it as white/gold on one site and blue/black on another, just because of the different colours the site had around the photo. It's really interesting to experience what context does to your perception, and it also underscores just how unreliable and subjective our perception really is. Well, yours is less so, apparently.

That's interesting.

The left ones look red to me when I look at the whole picture and orange when I look at one chip in isolation.

However, the right ones are definitely purple no matter what. Do people really see them as red?

Maybe I need to spend more time in violet lighting, I find this kind of lighting extremely alien and bizarre. For example, some reds turn into a dark, dirty shade of green.

I don't think I'm colorblind, btw :)

well, at first i saw red, then i read these comments and saw purple, now i don't know what to believe.

They look purple and orange to me, and I am not colorblind.

I have sometimes disagreed with other people whether a color is orange or red so maybe there's some other variation in color perception that's never been worthy of diagnosis.

The more I look at the three snails, the more I find A and C to be similar in their pattern of illumination.

EDIT: the closer I look the more I'm convinced. C is the matte rendering of A; B is the different environment map. Figure 2 in the paper bears me out; I think the authors managed to fool themselves while editing :(.

EDIT#2: make that figure 2A & 2B that bear me out ... 2C they seem to have used the same image twice in the bottom row. Am I mad? I know these errors are easy to make in the rush to publication, but my senses disagreeing with the text in so many places is starting to freak me out!

I came here to make the same comment. Overlaying the images with crossed eyes, you can easily see that snails A and C share the most similar pixels. The mistake is in Figure 1 of the original paper and has been copied to the article.

Except you have to look at it a long time and know that you're being tricked before you perceive it.

If the paper makes these kinds of errors, how can we be sure they didn't make similar errors when conducting the experiment?

Really, for this kind of paper, the only reasonable approach would be to figure out how to get actual babies to proofread it. Tricky.

No need to guess! You're right, A and C are closest - (A-C) shows fewer differences than (B-C). Here are the diffs for the images (with C-C as a sanity check):

http://i.imgur.com/GQGLE61.jpg

As other commenters have said, C is just the highlight-free version of A, which the diffs fairly clearly show.

EDIT: For reference, here's A-B.

http://i.imgur.com/DW8TL7A.jpg

There is a very interesting learn-to-draw book, "Drawing on the Right Side of the Brain", which takes a similar tack. Its premise is the reason people find drawing realistic scenes hard is because your mind is busy categorizing what it sees in to nice big-picture abstractions like "house" or "eye". You need to re-focus you mind to be able to see what is in front of you (how the lines converge, how shading affects shape, etc.), and once you do that, it becomes easy to draw in a realistic style, and not have your drawings come out like they did when you were 8 years old. Some of the exercises in the book are to copy line drawings that are turned upside down. This is so that it is easier for your mind to ignore the high level concepts, and just be able to copy down the lines as they are. I was certainly impressed and I wouldn't have guessed that by reading just one book and working the exercises I could have learned to draw so well. Highly recommended.

https://www.google.com/search?q=drawing+on+the+right+side+of...

https://www.google.com/?gws_rd=ssl#q=site:news.ycombinator.c...

On a throwaway account here. I'm currently teaching myself to draw using that book and I think the author is really onto something.

My first experience with psilocybin is also something that led me down a similar line of thinking. When the drug really hit me and I looked at anything visually appealing (a painting or a movie) all I could think was, "So this is how a kid sees the world!" A tree (I still knew it was a tree deep down) just looked different. I saw all the little parts that made up a tree. There was no longer a filter to intervene and say, "Yeah that's a tree. You've seen a million." It was like my brain did a revert to its state years ago and it was amazing.

Yeah, I’ve taken mushrooms, and my memories of being a young child are in retrospect very psychedelic, minus the visuals. (I imagine visuals are present at very young ages, say 0–1 years, before you learn how to fully separate your senses.) You take things at face value, it’s easy to lose context and permanence, you draw connections between things, and you feel insightful.

For example, I remember being about 3 years old, and realising that other people couldn’t hear my thoughts, and that I couldn’t hear theirs—it was mind-blowing. Of course, you tend to come to slightly more complex realisations while taking psychedelics as an adult, and therein lies the value: it’s a kind of renewal of your personality.

This is kinda the same thing as the himbu experiments - about how language and cultural taxonomy cause us to bucket things and then compare them on higher order dimensions.

With practice, you can turn this mechanism off and with sufficient focus and sincerity of purpose, see things without such taxonomy.

The extension of this exercise is when you are able to reconstitute things in different ways without necessarily any known categorical words.

De then re pattern things.

I think you'll find "language and cultural taxonomy cause us to bucket things" is mostly/all bunk.

The whole Eskimos have xxx words for snow BS goes along these lines as well.

http://languagelog.ldc.upenn.edu/nll/?p=17970

Are you making a counter claim that every language has the exact same number of words split the exact same way and that cultural context has absolutely zero bearing whatsoever in how events are seen or interpreted?

Yep. The essence in what you said.

Languages have different amount for the words snow, but it's not relevant to time in the environment AFAIK.

As long as you have snow, you'll be exposed to all forms hence similar amounts of names. England or Arctic. The increased exposure for the Inuit does not make for more words.

> cultural context has absolutely zero bearing whatsoever in how events are seen or interpreted.

As long as seen means by sight, and interpreted means lower level brain. And by zero you mean insignificant (Or way way less than these theories state.

I think if you research it more, you'll be surprised.

Universalism vs relativism.

What makes logical sense?

Language is used to describe the world around us seems very logical to me, it came second for starters.

If you want to go the other way, language shapes the world around us, actual evidence is needed for me.

It is full of "just so stories" which I find troubling. It's a meme that can spread without proper evidence beating a boring "language just describes things".

>The baby looks at things all day without winking; that is because his eyes are not focused on any particular object. He goes without knowing where he is going, and stops without knowing what he is doing. He merges himself within the surroundings and moves along with it. These are the principles of mental hygiene.

-Chuang Tzu

I wonder what impact this will have in creating toys for babies in that age range. Even if for marketing optics.

None what so ever, as it is the parents who purchase the toys... They won't notice any difference.

With the right branding, dog owners by balls that they know their dogs can see.

That and DOGTV.

It's weird to call it an ability of babies to perform well in a very contrived task, but perform worse in any real life object recognition.

It is the other way round. We adults have a great capability to identify similar real life objects in various lighting conditions. Hints like the reflections on the snail tell our brain about the lighting conditions and we automatically correct for it.

There's a bizarre but very fashionable form of self-loathing in which any way in which we differ from someone else, it is automatically assumed that the other thing is objectively superior. Do we, as adults, see one way but babies see another? Babies see objectively better! Does another culture eat differently than yours? Other culture is objectively better! Does the other culture have different music, art, story structure preferences? The article about how their way is objectively better than mine is already templated and ready to go in the minds of people held in thrall by this fashion.

I have to warn you that once you start seeing this, you'll see it in a surprisingly large number of places.

Seems strange to phrase this as something that adults can no longer see. If anything, it just shows that adult brains are more sophisticated in interpreting visual information.

> Seems strange to phrase this as something that adults can no longer see.

Its factually accurate.

> If anything, it just shows that adult brains are more sophisticated in interpreting visual information.

"Interpreting" is mostly throwing out the parts that are less likely to be relevant and keeping the parts that are more likely to be relevant. Obviously, this is a good thing on balance, and nothing in the article suggests that it is not.

It also involves not being able to as easily recognize certain difference (which are less likely to be important in most practical cases, but nevertheless are real differences.)

They've no prejudice and they question every status quo;

Reply on news.ycombinator.com