Rats in Doom VR | Hacker News Reader

Rats in Doom VR

medium.com

114 pointsdgellow52 comments

They seem to all walk in a straight line. Are they not able to make turns? It’s hard to tell if they’re even perceiving the VR environment or just doing their own thing on the sphere.

I suspect that TFA's author is aware of this, but rats' vision is so poor that I'm not certain VR is the correct modality for them. IIRC they rely on smell, primarily, along with touch, to navigate their environment. I imagine that the rat just decided to accept the likely-ineffective vagaries of the experimental setup, because the smell of food was tantalizingly nearby.

Did not Feynman use rat experiments in his book to lay heavily into social studies? As i recall in the book, there was only one guy whose rat/mouse experiments had enough rigour, eliminating data disturbing experiment influences, systematically.

https://calteches.library.caltech.edu/51/2/CargoCult.htm search for "All experiments in psychology are not of this type"

A paper found that in order to prevent rats cheating the floor needed to have sand to prevent them picking up on vibrations. The result should've been that experiments began incorporating this measure in their own setups, but instead the paper went ignored

The paper: https://psycnet.apa.org/record/1939-01955-001

Amusingly this VR direction could also be a useful method to limit the rat's perceptions to the experiment. But first it'd need to have the rat trained to explore a 3d environment & be fed by collecting food in the 3d environment

Well, we have better psychology experiments today. Just not published.

Run a filter for a situation and a query against a behavioural database at Amazon, Google, FB and you can build quite a nice insight into humanity.

Papers published with largest knowledge on humanity on earth : 0

The moral of that story, IMO, is not "don't do experiments with rodents" or even "bio/psych is nonsense."

Instead, the point is that a good scientist has to aggressively consider possible explanations for your data and possible experimental confounds. This is much, much harder when studying subjects that have some agency of their own, like rats or humans.

It gets even worse outside of science. I'm sure most people who've been in the business world long have noticed that attempts to eliminate confounding factors when drawing conclusions are rarely even attempted unless the results contradict what the highest-on-the-org-chart interested party wants it to be (and even then, they're usually just ignored rather than any attack on their validity actually being made).

For all the attempts to measure stuff and be "scientific", businesses mostly run on gut feelings, even when they're doing "experiments" or "collecting data in order to make informed decisions" or whatever. Often it's really easy to spot problems with these processes, but no-one cares as long as the PowerPoint slides say what they want them to say and pretend to be backed by data.

>but no-one cares as long as the PowerPoint slides say what they want them to say and pretend to be backed by data

I've seen this described as "decision-based evidence-making".

VR has become a pretty popular tool in neuroscience.

Rodents can be trained to follow cues shown in VR, or even navigate to specific virtual locations by running on the sphere, so they can certainly perceive something.

That said, vision is certainly not their primary sense and the organization of their visual system is...dissimilar from humans and other primates.

Calling it VR sounds like a stretch. I doubt the mouse is experiencing immersion instead of just reacting to some blobs of color on the screen.

Eh, it’s partly marketing, but there are some neat tricks that make it more immersive.

Some groups put the rodent at the center of a translucent hemisphere, then reverse-project the scene onto it. This can completely fill its field of view, and looks pretty good if you get the transform right. It’s also not too hard to make walking on the ball control the camera, so that objects approach faster when the animal runs, or the scene swivels as the animal turns. This article has a few videos of the setup: https://www.nature.com/articles/d41586-019-00791-w

I doubt the mouse is totally fooled (but neither are you with an Oculus). I think someone has done some cross-over experiments: mazes learned virtually transfer over to physical contraptions and vice versa. Unfortunately, I’m blanking on who might have done this.

From the article: "Although the mechanisms to train left-right turning was present, I did not have time to train movements with turns included, so in most sessions only one axis of motion was registered."

To folks saying this is heartless, I'd urge you to scroll down to the "Experiment" section of the article:

> I received three 8-week-old male Long Evans rats, Romero, Carmack and Tom. Romero was fearless (more like thrill-seeking) and loved grapes. Carmack was a real architect building around its home keeping it tidy; he was fond of bananas. Tom began shy, but held the most surprises in learning performance.

> I spent around 6 weeks with them during their (and my) night cycle, ~1 hour with each daily. It took two weeks to habituate them to me — this process could be accelerated, I was just a little slow with them. One week to teach them to stay put while I was dressing them; I also tested multiple harnesses and their combinations during this time.

It seems to me from this that the rats were well-cared for and the author formed a bond with them, considering how he seems to fondly recount their personality quirks.

That doesn't really reassure me.

I don't know anywhere near enough about rats in general or this particular situation to say whether or not this is something bad for the rats, but this recounting of basic bonding doesn't really tell me anything.

Just anecdotally, I had a pet rat as a kid. They're pretty friendly creatures who seem to enjoy company and playtime. Not sure what constitutes "bad for the rats", but I personally think a situation where they're getting food, safety and stimulation would not fit into that category.

Regarding playtime, rats love to be tickled. There are many videos available out there showing it. It really opened my eyes to what was going on in a rat's brain that they would enjoy "pretend goring" by their caretaker and found it enjoyable.

I think the final line is more reassuring:

> The experiment was approved by the Feinstein Institutes’ Institutional Animal Care and Use Committee.

Sounds like he went through the proper channels.

From the videos, I got the impression that they just followed a straight line trying to reach the food that was in front of them.

thats all that was shown, straight line travel, doors reward with food, running into an enemy blocks travel until enemy dispatched.

I wasnt impressed until i realized this id a starting point and it could be further developed potentially.

Despite all the gloom I am still just a rat playing Doom

then someone will say what is lost, can never be saved.

That's just mean. Those poor animals how creepy can humans get. Leave them alone.

This is a major improvement over how these things were done in the past, which often involved applying stimuli over brain diodes.

If you think this is bad you should see what they do to lab rats >.>

Also sad and often unnecessary, but at least there’s usually some legitimate science instead of whatever this is.

The rats applied for an internship

How is it mean? They get all their needs provided for, two companions, no fear of predators, and every so often they get put out on a ball which dispenses candy when they run.

Reading the article, it seems like this was exactly their situation. It seemed like it was in the author's best interest to keep them comfortable and friendly since this was an operant conditioning task. He also mentions the only negative reinforcement considered was a puff of air, but he opted not to use it.

I'm pretty sure this is exactly how Cave Johnson got started at Aperture Science...

https://half-life.fandom.com/wiki/Cave_Johnson

Nah, I think they were making shower curtains.

Does anyone notice how he used a trackball mouse design to track their motion in VR?

I feel like there is a clever joke in there somewhere.

Just reinforcing the stereotype that a rat is just a giant mouse

My thought was mainly "ok how can I build a ball big enough for a VR rig for myself?"

After this and the recent conversation about the 'rat shadows' setting in Dishonored, I just keep humming to myself-

♫♪♪ The last that ever she saw him ♫♪♪

♫♪♪ Carried away by the Doom-rat shadow ♫♪♪

Ahh, this post feels like it's gonna be a footnote in the history of the Meta Matrix.

Why don't we just train rats to react to transpiled digital stimuli instead of CNNs?

This should be called Reinforcement Learning :)

Can they navigate turns, or are they just being conditioned to run in a straight line for sugar?

Interesting exploration though.

This is insanely heartless.

That's a little strong. Arguably a bit inhumane but nothing like what we put them through for drug tests I'm sure.

Do animals (incl. humans) who go through one sort of torture find comfort in the fact that there are worst forms of torture?

Can you elaborate on what about this is torturous? It sounds like the experimenter treats the rats well, got them acclimated to his presence/the harness/the treadmill over time, and only ran the tests with them for less than two weeks.

Well this doesn't look like "torture" to me for a start. Admittedly I don't have much knowledge of what goes on in a rat's mind but this looks uncomfortable at worst. He says in the final line: "The experiment was approved by the Feinstein Institutes’ Institutional Animal Care and Use Committee."

Clearly I don't think that simply because worse torture exists that somehow placates a rat's mind, but as a human I choose to point my outrage where it is most effective.

Also, aren't these lab rats? Where would they be had he not picked them for this experiment?

Sometimes feel like _I'm_ a rat in a maze of doom.

cruel, leave them alone

Times have changed, lab rats are now playing a VR video game.

seems cruel to me.

You would really hate to see the mouse house and depending labs of any research hospital then. The mice here are living a life of luxury by comparison. Yeah, yeah, I know you or someone else will come out against those practices as well; I would love to hear ideas for a replacement or a proposal to just stop science and progress all together.

I don’t have a strong opinion for or against the use of rodents in science, but the tone of your comment was uncalled for, in response to someone simply expressing compassion for a living creature.

But that’s medical research. This is just a hobby project.

It seems completely reasonable to object to using mammals in one’s tech projects, while admitting the value to using animal models to advance medical science.

There are loads of smart people advocating for alternatives to animal testing.

Easy example from Johns Hopkins: the center for alternatives to animal testing https://caat.jhsph.edu/

Can you play against the rats?

Reply on news.ycombinator.com