Ultra-Rapid Vision in Birds | Hacker News Reader

Ultra-Rapid Vision in Birds

journals.plos.org

65 pointsdownboots19 comments

Hmm. It is possible to distinguish flickering lights at much higher than the flicker fusion threshold by using eye motion or other types of fast motion. A constant light will produce a smooth blur under motion while a flickering light will produce many distinct images. It seems like they don't have a way of checking whether the birds are using motion to distinguish the flickering light or not.

I would go one step farther: I bet that plenty of humans could pass the test that the birds were subjected to at 120Hz or even higher. They didn't test whether the birds could actually resolve stimuli at high temporal resolution -- they tested whether the birds could detect flicker. A bright source pulsed at 120Hz is easily perceptible (and incredibly annoying), especially if it's a square wave with a duty cycle that isn't especially high.

Where are the control animals?

I don't think it's true that 120 Hz flicker is easily perceptible, absent motion. Flicker fusion is real. But motion is common, so that's why humans in practice can detect 120 Hz flicker (and find it annoying. Ban PWM taillights).

PWM itself is fine!!! I have no idea why they set the PWM frequency so low. Even setting it to something like incredibly low like 1kHz would have solved the problem. In electronics land you almost have to try to get frequencies that low out of e.g. a microcontroller - you need a very high clock divide ratio to get a timer PWM period that long.

I think they have actually done so, because I am noticing fewer low-frequency taillights these days.

From vague memory of reading some data sheets: there are cheap little constant current LED driver modules for automotive applications. Two wires in: PWM power. Two wires out: the LED array. To get anything less than full output, you need to drive it with a square wave, and it might not function at a civilized frequency of a few kHz.

Doing better would require a different wiring design — there’s no way to just swap the driver without making the driver fancy enough to take, say, 50% PWM in and produce half current DC out. (Obviously this is trivial, and even available entirely off the shelf for non-automotive applications, if you have three wires in. But you don’t.)

Watch a falcon pluck a drone out of the air, inserting its talons between the blades. They see everything faster.

When driving on highway, the wheels of other cars looks smudged. If i flicker my eyelids i see a snapshot of an other car's wheel like if it were not rotating, i.e. a strobe light like effect. That is ~1500rpm for human eye (and for us it is about 30ms for the signal to pass through the first stages of the visual cortex). Birds are several times faster (even if just for the much shorter physical path from eyes to the rest of the brain and of the related paths inside the brain).

Actually, a semi truck tire is about 127" around. At 60mph, 1m per minute, that is more like 500rpm. For a car tire (80") that would be a little under 800rpm at 60mph. So you are seeing wheels spin aboit 10x per second, which is well within normal perception if you concentrate. Paint a dot on the tire and you would see it spinning around.

Drone blades spin in the thousands, high thousands of rpms.

This is an inherent limitation in trying to use Hz to measure a system which does not really have a concept of frequency at all. For instance when you have a motor system -> vision feedback loop, humans have been shown to accurately discern latency differences of as low as 1ms in some studies. Though if you've ever played on a first person game at 60Hz, then 120Hz, then finally 144Hz and beyond, this probably won't surprise you. It's quite strange how you can perceive multiple degrees of "instantaneous".

It's also been shown that latency differences as low as that don't really have any noticeable impact on human performance though, so it's likely we can merely perceive that to allow our brain to subconsciously fine-tune our motor system. You'd be a very clumsy human if your motor system only had a resolution of ~20ms throughout. Despite it obviously being necessary to help you learn to use your motor system, we don't really seem to get to use that high "resolution" much consciously.

Also I might be comparing apples to oranges here, because you could also argue that a camera taking one picture every 10s could discern differences as low as that, if you take the pictures at the right time. But we also don't work like cameras, which brings us back to the topic at hand of frequency not being a good metric since our vision is more of a continuously operating system.

Make of that what you will.

Flicker fusion studies typically control for this motion-based detection artifact by using head-fixed preparations or tracking eye movements, which the authors here attempted by using stationary perches and monitoring head movements during trials.

As far as I can see the paper doesn't mention doing this at all. Do you have a secret source or did you just make it up?

Echoing to this perhaps, I heard birds can react 13 times faster than we do, which is especially useful in flight with hundreds or thousands of other birds.

you should see the reaction time of the fly that's been in my house for days now

The trick to catching a fly is to move very slowly

Correct, or let your cat do it :D

I have heard tell that this works. I have it on somewhat good authority that old ladies may also work with unpredictable results.

https://en.wikipedia.org/wiki/There_Was_an_Old_Lady_Who_Swal...

Wow this brought back a hazy memory of school sing-a-longs, thanks.

Yes, smaller animals, faster heartrate, more direct wiring, they probably experience time different than we do.

Many human responses are purely unconcious muscle memory due to nervous system latency, this also implies the brain has evolved to be highly predictive as to compensate.

> Many human responses are purely unconcious muscle memory due to nervous system latency, this also implies the brain has evolved to be highly predictive as to compensate.

Which I’m reminded of every time I reach to grab a knife I’ve dropped, instead of just stepping back and letting it fall.

Reply on news.ycombinator.com