Naked Science Forum
Life Sciences => Plant Sciences, Zoology & Evolution => Topic started by: jfoldbar on 08/07/2021 20:27:33
-
im not sure if this is science or philosophy. please let me know if i should ask elsewhere.
i am having a debate with a friend (real life) about beliefs. and we got stuck on whether animals have beliefs or not.
so i would like peoples input on this subject.
-
I'm going to say yes, at least in a general sense. I recall that pigeons (or some similar birds) were taught to have superstitious beliefs by randomly being awarded food. They developed a tendency to associate whatever behavior they were performing at the given time of reward with the appearance of the reward. So they tended to repeat the behavior as if hoping it would trigger further rewards.
-
thanks for the reply kryptid. im assuming you are referring to the skinner experiments of the 50's.
i was thinking about this, and i am not entirely convinced this could definitively translate to a belief, for a few reasons.
i will try to explain.
1, if the pigeon believes that spinning around gets him some food, does an ant actually believe that he can eat your foot. how many ants does the colony need to get that food? the pigeon problem should also extend to other animals smaller and smaller. does a germ exhibitor some kind of behavior that could be considered a belief or superstition?
2, imagine you have a robotic pigeon. some super advanced one from the future. and this pigeon is nothing but computers and mechanical parts. if skinner done his experiment on these pigeons, it may well be that one of the pigeons exhibits a similar behavior. why would a robot spin around to try to get the food? because it appeared to work previously? so it has "incorrectly" learned a behavior and without more data or input has decided that this is a chance to get food. eventually it could calculate (if its mathematical algorithms are advanced enough) that there is no correlation between spinning around and food, but it would need time to do so. however the real pigeons could eventually learn this too.
my point is, in my opinion, if animals have beliefs then so too would robots. which i assume most people would think, they dont.
i realize this is becoming so far into the philosophical realm. but i find it an interesting topic.
-
My dog has a belief in gravity.
If he sees me put a treat in a high place, but when he looks, its not there, he immediately goes sniffing around on the ground.
He also has a belief in the permanence of matter - if it's not there, it must be somewhere else.
The same sorts of experiments have been done on human babies - put a toy in a box. The toy is then (sometimes) removed via a hidden trap-door. When the toy disappears, the baby stares at this unexpected event for longer than if the toy is still there (as expected).
The purpose of a brain and neurons is to predict the future, given the current events as inputs. This implicitly encodes beliefs and expectations in the connectivity and excitability of interconnected neurons.
This same structure is imitated in artificial neural networks, and in the machine learning algorithms that train them.
- If you train them with nonsense inputs, you will get nonsense outputs: "Garbage In, Garbage Out"
- If you train them to keep eyes on web pages, you will train them to feed you ever more extreme and polarized hate speech (like all the social media do...)
So, yes, robots will develop beliefs, too. Whether they could enunciate their beliefs is another thing; there are projects at the moment looking at "Explainable artificial intelligence" - you can ask "Why did you say that?", and hope to get a half-intelligible answer.
See: https://en.wikipedia.org/wiki/Explainable_artificial_intelligence
-
If you define belief as accepting a hypothesis in the absence of evidence, it is highly likely that all intelligent species do it to some extent.
Recognition of temporal, spatial or other sensory patterns is essential to survival. But the world abounds with false patterns and partial patterns.
Several innocuous insects have evolved wasp-like stripes that apparently protect them from visual predators. This can only work if the prospective bird believes that yellow stripes are always a Bad Thing - a safe hypothesis but not supported by evidence if you actually eat a hoverfly.
Partial patterns or patterns buried in noise are just as important. A dog detects thousands of trace scents but decides that there is "rabbit" buried in the olfactory noise, and sets off hunting on the basis of partial evidence. If the trail breaks, he believes that the rabbit is still somewhere in the vicinity so starts a widening grid search, accepting the hypothesis that rabbits can't fly.
A simple GPS road navigator uses a predictive algorithm to display your estimated time of arrival based on actual progress so far and the hypothesis that the traffic ahead is not congested. A slightly more sophisticated GPS receives actual traffic information and a good airborne GPS uses broadcast actual and forecast wind to give you an extremely credible ETA based on the hypothesis that your airspeed will remain as planned. All reasonable hypotheses but definitely in the realm of belief because by the time you actually know, it's all over!
How far can we extend the beliefs of machines? A wise driving instructor told me that you should continue to drive towards a dog in the road because they have the wit to get out of the way, and you don't want to be rearended by the bloke behind you who hasn't seen the dog. But a cow probably won't move, and a bull or a ram may even decide to charge the car, so slow down or stop because the bloke behind will at least see your brake lights and a nudge from behind is better than hitting a ton of beef at 40 mph. Can a self-driving car acquire enough information to distinguish between a large dog and a small cow?
The comes the anticipatory hypothesis. Parked cars either side of the road at 4 am: very little chance of anyone suddenly running in front of you. 4 pm, almost a certainty.
-
well. a lot to digest here. first i would like to express my thanks for the logical detailed responses. i have actually been trying this topic on reddit for about 6 weeks now, but the mods kept deleting it. (wont allow this subject). so i am glad to be getting replies like this.
evan_au, you say that robots will develop beliefs too. i have trouble imagining that, but i guess plausible.
the problem i have is it is so hard to say is there a line between belief and calculation. (for robots and for biology)
if i may, i will explain what i mean in relation to terminator, because its something that 99% of the population could relate to.
terminators programming will have things like jumping in front of a truck may or will result in damage. shooting at the policman for little reason may result in 10 policeman shooting back. potential for damage. however a fast cpu could be calculating the risks for how much potential damage verse how much is achieved, hence why he could storm the cop shop in T1. he can calculate that there is low chance of those weapons hurting him.
is this a belief? or a calculation?
if he was chasing sarah conner, and she ran across a glass roof strong enough to support her, would he chase? what if there was lava under the glass? would that change his plan to perhaps go another way. what if the lava is only 1 foot thick, and he doesnt see another way.?
so he runs across, the glass breaks and by the time he gets out of the lava, he is 1 foot shorter, running on stumps.
does he believe his legs are gone, or is this a calculation?
this concept has no end to it?
does the cpu in your car believe that its driving the car, or does it know it?
as i said last post, getting into philosophy.
alancalverd.....
i know what you mean about the dog. i had this same discussion with my friend.
however, is this actually a belief? (so much philosophy here it hurts my brain).
the dog may believe the rabbit is in that hole, does a mozzie believe my blood tastes better than yours? does a microbe believe its food is that way? hard to say where the line is?
does the dog believe its tail is part of it, or something else?
how about a dog that chew on their own leg until its bleeding, thinking its a bone?
this could go further.
do i believe that i have 2 arms and 2 legs? or is it a fact?
halc..
self preservation may not be belief though, it may be just a result of numerous biological processes whos need is for self replication.
germs self replicate, however they probably dont know that they are doing it. but there is still the inbuilt biomechanical need to. do ants realize they are self replicating? if they dont, why build a house that helps them do this?
again, the line is so fuzzy, it hurts my brain.
-
I would say definitely, but it of course depends on your definition of 'belief'.
Any creature complex enough for a sense of self preservation requires a belief in the self in order to be fit.
This means bugs, amphibians, etc. Probably not sponges.
evan_au, you say that robots will develop beliefs too. i have trouble imagining that, but i guess plausible.
the problem i have is it is so hard to say is there a line between belief and calculation.
It would seem that a belief (at least an acquired one) might be the result of a calculation, for both a biological thing or a robot. My beliefs seem partly to be a result of my calculations, but many of them are not acquired by calculation, but rather bred in, making them the result of DNA in the case of a biological being, essentially equivalent to hardware programming, not beliefs acquired by running 'software'.
if i may, i will explain what i mean in relation to terminator
Not me. I have a hard time making scientific analysis of a fictional anthropomorphic robot, which seems as useful as discussing the physics of Back to the Future.
I'd rather discuss something like the automatic lawn-mowing robots (the grass version of a Roomba), or a self-driving car. These things are real.
does the cpu in your car believe that its driving the car, or does it know it?
The CPU is just a mechanical device, like a brain cell. It is involved in calculations, but it is the entire system working as a whole that accomplishes the task. I believe electricity would hurt me. No particular brain cell believes that. But it was learned, so it is a result of calculation. My belief in my 'self' on the other hand is hard-wired in by hundreds of millions of years of evolution.
I seem to have multiple parts composing my full consciousness. Some of those parts hold contradictory beliefs, so my end belief might be due to which part of me is in charge. This sort of introspection has exposed some interesting facts to which I see few references on the web. Maybe I just don't know where to look.
does a microbe believe its food is that way? hard to say where the line is?
Hard to draw the line indeed since it is probably best expressed as a spectrum, not a binary thing. It seems that the microbe is fully reactional and doesn't require beliefs to aid in decisions.
do i believe that i have 2 arms and 2 legs? or is it a fact?
Debatable, but I doubt too many creatures other than humans spend too much time on pondering the factualness of such subjects.
self preservation may not be belief though
No? I believe I will die if I jump off this cliff. At least the part of me that's in charge believes that. You have to reach to very primitive creatures to find one lacking in such programming.
it may be just a result of numerous biological processes whos need is for self replication.
Oh, it is very much a result of numerous biological processes, but how does that make it not a belief?
do ants realize they are self replicating? if they dont, why build a house that helps them do this?
I doubt ants realize such things, but they still avoid damage to the extent that they detect danger. The thin line might be between instinct and belief. The former builds the house and assigns the tasks to the workers. Where does belief fit in with the ants? Again, it seems more primitive than the instincts. Nothing complex comes from the beliefs. If I swat a mosquito and miss, it flies off in reaction, and quickly lands again. After 12 such failed attempts, do you think the mosquito believes there is a conspiracy to kill it? Probably not. That's a reasonable argument that the flight reaction is instinct more than a belief.
Just trying to argue both sides. Not pretending to know the answers.
-
thanks for the reply kryptid. im assuming you are referring to the skinner experiments of the 50's.
i was thinking about this, and i am not entirely convinced this could definitively translate to a belief, for a few reasons.
i will try to explain.
1, if the pigeon believes that spinning around gets him some food,
I think there are a number of researchers who have questioned Skinner’s conclusions. One is that the pigeons were experimenting to see what worked as they appeared to try a number of behaviours. That probably indicates a belief in cause and effect.
I have seen video of animal experiments with mirrors which seem to indicate that some animals have an understanding of self and not self. Pigs and dolphins have it apparently, can’t remember the conclusion on dogs.
Difficult area to study and remain fully objective, but interestingly discussion here.
-
If a belief is defined by thinking something is true, even when there is no evidence, then how does an animal get "told" that the thing is true.
The pigeons have got evidence- their recollection that "It worked before"- is evidence, albeit not very strong.
To have a belief in that sense, there has to me a means by which they learn to do something, but not from experience (otherwise it's "just learning").
-
To have a belief in that sense, there has to me a means by which they learn to do something, but not from experience (otherwise it's "just learning").
This is why I have a problem with the idea that pigeons might be superstitious based on the experiments. To me a superstition is a belief based not on experience of cause & effect, but on an irrational connection between 2 events. If I walk under a ladder and someone working on it drops something on me then I have a valid reason for being wary of walking under ladders. To believe that there is a connection between breaking the trinity = walking under ladder and bad luck is quite abstract and requires some rather convoluted thinking rather than just trying things out and repeating what works.
-
you say that robots will develop beliefs too.
Today's AI systems are trained "offline" on powerful computer farms, fed by huge volumes of pre-recorded training data.
- The optimised neural network is then loaded into a much less powerful processor for execution (but far more portable and economical to manufacture and run).
- So today's AI systems don't "learn on the job", they are pre-programmed.
- The best you could hope for today is to capture a case where the AI system did something stupid, and add that to the training data as an example of what not to do.
- But one day, AI systems will become more self-training, and I expect they will develop peculiar beliefs of their own, based on their own unique training inputs.
I heard of a story where a self-driving car, every time it approached a certain bridge, "believed" the road went in a different direction from what the road engineer designed.
- The driver had complained about it
- And one day, the driver didn't notice that bridge was coming up...
See: https://abcnews.go.com/Business/tesla-autopilot-steered-driver-barrier-fatal-crash-ntsb/story?id=68936725
do i believe that i have 2 arms and 2 legs? or is it a fact?
I believe that the vast majority of people have more than the average number of legs.