This morning I found myself wondering the minimal threshold of cognitive sophistication to be subject to logical fallacy.
In our household, we have a morning routine in which our 2 cats start yowling to be fed as soon as it begins to get light outside. Because this is so annoying when we’re trying to sleep in, we make a concerted effort to “untrain” the behavior by not feeding them when they start yowling. We don’t feed them until we’ve gone through both of us coming downstairs, retrieving the newspaper, and brewing coffee (all at 6:45am or so). We never cave into their demands to be fed at 5:00 or 5:30am.
In spite of all this, they never learned that it’s pointless to set up a racket at first light.
If I were a human, I’d say they were falling prey to a cause and effect fallacy, or being deceived by regression to the mean. We eventually get up after they start screaming (because it’s daytime, and eventually we must get up). Thus the behavior is reinforced, even though our response of rising is a good hour delayed from the onset of their stimulus of howling like they’re being murdered.
On thinking about it some more, I realized… they’re just following a stimulus/response sequence with no opportunity to falsify the pattern, so why would it ever be otherwise? If it were a network of rat neurons in a petri dish, wouldn’t that neural net eventually train to the same fallacious conclusion? Sun rises, therefore begin annoying behavior, and food comes afterward.
I’m not sure where I’m going with this other than the apparent conclusion that poor reasoning seems an inevitable part of any nontrivial neural system. Thoughts?
I read somewhere (no cite) about an experiment with birds (chickens? pigeons?) where there was a repeated delivery of food made following a particular action of the bird, which was subsequently randomised, but the birds continued to repeat the action ad infinitum despite the cause-and-effect being randomised. I’m guessing even insects could be persuaded into erring in this way too.
My anecdotal observations of flawed kitty logic: my ex wife used to use her phone as an alarm clock. It had the vibrate function on, and would move around on the bedside table as it rang, before she got up and fed the cat. The cat therefore concluded “this object moves, I get food” and would sit on the bedside table and bat the phone about, then look around expectantly for the food. Similarly we used to put a collar on her before letting her out in the garden, and she associated the box the collar was kept in with her escape from the house, culminating in her knocking the box off the shelf so that it spilled the collar out onto the floor. She would then look around, confused about why she wasn’t yet in the garden.
Since my ex moved out, though, I retrained her to accept my morning ritual happening before asking for food, and she no longer stomps all over me at 5.30 - so even dumbasses like her can learn.
No cite either, but I read that animals will do this when there is no correlation between stimulus and feeding to begin with. They’ll just pick a ritual and keep repeating it, when left alone in a cage with a random feeding schedule. Scratching the floor three times with the left foot, pecking at the cage in a particular way, etc. Even more interesting is that humans will do it too. Sit someone on a chair with a display and some buttons, and at random times the “score” on the display goes up. Can you guess what they start doing?
It was actually a little more than that. It was a famous experiment by BF Skinner done in 1948. The birds were not first trained to respond to a cause and effect stimulus. Instead, they were made to be very hungry by reducing their food, and were given treats at completely random intervals. What happened was they developed ritualized actions (bob three times, turn around 2 times, step once to the left, etc) designed to “prompt” the random occurance. The birds repeated their rituals over and over once they became associated with success by an accidental food drop. Skinner described the birds’s actions as “superstitious.”
I don’t see any poor reasoning, just simple reasoning. You might argue that their noise is a waste of energy because you will feed them at 6:45 regardless, but that is ignoring a billion years of evolution in which the situation might have been different.
Do they actually have enough information to know that making the noise is not going to gain them anything? Not sure.
It’s not an inevitable part of any non-trivial neural system, but it might be an inevitable part of one configured through evolution within the environment organisms have lived in here on earth.
I think the issue is how decisions are made with limited information. The cats don’t really understand your situation. Most mornings you’re up much earlier and they get fed when the sun comes up, but every once in a while, you don’t get up right away and don’t feed them. They don’t understand work weeks vs weekends or alarm clocks.
We can laugh and say “Oh, those cats are so simple and mistaken” but no cognitive system (other than God) is going to have complete information. All we can do is take the information available and do the best our cognitive faculties allow for to assemble that information into something useful.
Conclusions of powerlessness and randomness are not very useful. In fact, if someone said “Doc, I think random factors are ruling my life and I have no control over them,” the doctor would be halfway to a diagnosis of depression right there. It’s much more useful to assume that things are not random and that you can exert influence over them. Even if you’re wrong, it keeps you more actively engaged with your environment and more able to take advantage of situations where it is true.
On the other hand, I’ve heard of another experiment (I think also by Skinner) where rats managed to outsmart humans by avoiding this sort of superstition. The subject is presented with two buttons, each with a light associated with it. The subject presses a button, and a light turns on. If the button chosen and the light match, then the subject gets an incentive (a bit of food for the rat, or probably money for the human).
Well, it turned out that which light actually came on was completely random, but not evenly split: The left light had an 80% chance, while the right light had a 20% chance. The rats fairly quickly learned to just always push the left button, but the humans kept on trying (and failing) to find patterns in the randomness, like “After three lefts in a row, it’s right”, or things like that.