About the Inferential System II

This continues our exploration of the Inferential System, using basic facts about perception.

A Single, Coherent Conclusion

Look at this figure.

This is called a Necker cube. If you have never seen it before, what you almost certainly "saw" (consciously experienced) a three-dimensional cube. But there are two possible orientations of the three-dimensional cube -- there are two squares in the drawing, and either one could be the front of the cube.

This then is your normal conscious experience: When something is ambiguous, you do not perceive both meanings. Instead, you perceive just one. Putting this in the language of the Inferential System, your Inferential System just presents one meaning to consciousness. Research suggests that your Inferential System contains, at least momentarily, both interpretations of an ambiguous figure. This then is one more way that your unconscious is alien -- consciousness contains just one coherent interpretation of the world. while your Inferential System can contain several.

Once you know the Necker cube is ambiguous, you can try to "see" both orientations. In other words, you could try to change which meaning your Inferential System presented to your consciousness. If you practice, you can learn to switch easily between the two orientations. But you would never consciously experienced both meanings at once. This then is a second basic fact about how the Inferential System works: When the sensory experience is ambiguous, the ambiguity does not naturally become conscious. Instead, the Inferential System presents only one meaningful conclusion to consciousness. This of course applies to any ambiguous sensory information, not just the Necker cube.

When I worked as a cashier in a restaurant, sometimes waitresses would come to me and ask for tape. I had learned from painful experience that this was ambiguous -- they could mean adding machine tape, or they could mean something like scotch tape. So I would ask what kind of tape they meant. But they didn't know about the ambiguity, so they became frustrated with me -- they just wanted the normal adding maching tape, they didn't know its name. Or they just wanted to stick something to the wall, and they didn't care if I gave them scotch tape or masking tape.

If I say "Come here," do I mean right now or just at some time? Do I mean right where I am standing, or near where I am standing, or just the same room? We usually do not notice the ambiguity because our Inferential System correct figures out the correct meaning and we just go on with our lives. For example, if I saw that the boy hit the ball with the bat, you do not worry whether bat is a piece of wood or an animal, and you are probably correct in just assuming that it was a baseball bat (in the U.S.) and a cricket bat in Great Britain. You do not worry about alternative meanings of hit or ball.

Guessing

You might not like the fact that your Inferential System guesses at the meaning of ambiguous things and does not tell you that it is guessing. You might be thinking, "I might make mistakes, thinking that I was receiving an accurate interpretation of the world when my Inferential System really was just guessing." Yes indeed, mistakes occur.

But if your Inferential System didn't guess, you wouldn't get much information out of it. We live in a world of unnoticed ambiguity. For example, the Necker cube could be a two-dimensional object, and I think there are hundreds of possibilities for three-dimensional orientations, though I can't see any of them.

Or suppose someone says "Can you stand here for a second?” You probably know roughly what is meant, but there is a lot of potential ambiguity. They could be asking if you have the ability to stand, though they probably are asking if you would stand there. They probably mean the immediate future, though they don’t say so. In any case, they don't exactly say when you should start. "Second" is probably not meant to be precise, so the duration you need to stand is ambiguous. "Here" is also ambiguous. They probably don't mean the exact spot they are in -- you in fact can't occupy the same place they are in. Perhaps they mean that you should stand in exactly the same spot after they move (as in setting up a camera photograph), or perhaps they mean just to stand nearby. (And that's just the plausible ambiguities.)

So, you live in an ambiguous world. If your Inferential System warned you when its interpretation was a guess, the warning bells would be sounding all the time. If your Inferential System didn't guess, nothing would become conscious. It really has no method of presenting the ambiguities, but if it could, your consciousness would be overwhelmed with possibilities. Anyway, your consciousness isn’t any good at figuring out which interpretation is correct -- your Inferential System is skilled at perception, not your consciousness. So, your Inferential System makes its best guess and presents just one coherent interpretation to consciousness, and that’s probably a good way for your brain to work. For example, it told you that the Necker cube was a three-dimensional object, though it could have been a collection of lines in only two dimensions. Did you really want to know about this ambiguity? Obliviousness to Guessing I once did an experiment in which people said "Sunday". They tried to sound either happy, sad, friendly, unfriendly, glad that Sunday was the answer, or disappointed that Sunday was the answer. It turns out that happiness, friendliness, and gladness all have the same high pitch; sadness, unfriendliness, and disappointment have the same low pitch.

I would ask if a friendly utterance sounded friendly, and the subjects would say yes; I would ask if the speaker sounded happy (for the same utterance) and they would say yes; I would ask if the speaker likes the fact that Sunday is the answer, and they would say yes. So, even though high pitch is ambiguous -- it could signal happiness, friendliness, or liking the answer -- subjects didn't let that ambiguity slow them down a bit. Instead, they freely the utterance it all different ways, depending on what I directed their attention towards.

Let he who is without sin cast the first stone. About this same time I met a woman I liked. I said hi to her, eager to find out if she liked me. No high pitch. I was very disappointed, thinking that she didn't like me. But of course the lack of high pitch was ambiguous; I just didn't think about the other possibilities. As she later explained to me, she didn't want her friends to know that she liked me. (We were later married, so I trust that she really did like me.)

Similarly, Chinese is a tone language. In tone languages, the pitch signals meaning, so speakers can't vary pitch just to signal friendliness. When Chinese people speak English, they carry over this habit and don't use pitch to signal friendliness. They "sound" unfriendly, but this apparent unfriendliness is just your Inferential System making a wrong guess.

Consistency

Here's a problem: Johnny's age equals the number of quarters in his pocket. How old is Johnny?

There is no way to know how old Johnny is. Not enough information was given. In our culture, we are accustomed to problems with just one right answer, where there is enough information to figure out the right answer. If I say that Fred is 5 feet tall and his age is twice his height, you can figure out that he is 10. This process is called deduction. The problem I gave you was "unfair", because it violated that rule.

The Inferential System doesn't work deductively. There isn't enough information to figure out which orientation of the Necker cube is correct, but your Inferential System still produced a conclusion. Because the world is ambiguous, there never is enough information to know anything for sure, but your Inferential System still produces conclusions. Its conclusion does not necessarily correct; it is just consistent with the given input.

Fortunately, your Inferential System doesn't make wild guesses, and in fact it usually chooses the most likely interpretation,. But it does guess. In fact, you probably have an image of Johnny in your mind, and you can tell me about how old he is in that image. That is your Inferential System's guess at Johnny's age. And it no doubt is a reasonable guess. Johnny could be 100, but that would be a lot of quarters; you probably imagined someone much younger. Johnny could also be 0, but you probably imagined someone old enough to have his own quarters in his own pocket. Of course, to form an image of Johnny and his quarters, your Inferential System had to include age. So your Inferential System made its best guess.

Automatic Use of Cues

As you probably know, the two lines in the following figure are the same size.

However, the first line looks longer. This is called the Muller-Lyer illusion. It confirms my model of perception. If people had conscious access to the visual perception of lines, they would know they were the same length. Instead, they are conscious only of a conclusion. In the Inferential System's conclusion, the first line is longer.

There is another curious issue about illusions. The first line looks longer because of the flanking lines. But why would anyone take the flanking lines into account when trying to judge which line is longest? It’s a foolish thing to do -- the flanking lines can’t help, and they could be (and are) misleading. But your Inferential System automatically uses the flanking lines as a (misleading) cue. Perceptual illusions, such as the Muller-Lyer illusion, are based on the automatic use of misleading information.

The point is, your Inferential System uses its cues automatically. Consciously, you cannot stop it from using a cue (and in fact you probably don’t know that it is using the cue.) This automatic use of cues probably occurred in the problem about Johnny and the quarters -- your image was probably influenced by the fact that I called him Johnny rather than John. And of course, it occurs all the time -- when you perceive a cat, a lake, or whatever.

In the case of illusions, your Inferential System is misled. But usually this automatic use of cues works out fine, because most cues are not misleading. EXAMPLE Anyway, it would be a complete mess if consciousness had to tell the Inferential System which cues to use. Consciousness doesn't have time to attend to all the cues and it wouldn't know which cues to use, to name two problems.

The Influence of Unconscious Sensory Input

Suppose something is presented very quickly, say on a computer screen. If the presentation is too short, the subject consciously sees nothing; if the presentation is long enough, the subject becomes conscious (sees) what is presented. But there is a small window in between these two possibilities, in which the presentation is too fast to be consciously seen, yet lasts long enough to influence the Inferential System. This is called subliminal perception.

Subliminal presentations can have any of the effects on the Inferential System that normal presentations can have. If subjects first perceive the word DOCTOR, this speeds up their perception of the word NURSE. This same effect can occur even when DOCTOR is presented subliminally. Or, as already discussed, the Inferential System uses cues to guess at the meaning of ambiguous words. Presenting TREE will then cause subjects to later interpret PALM as being a tree; presenting HAND will cause subjects to then interpret PALM as being part of the body. This effect can also be produced with subliminal presentations of TREE and HAND.

The right conditions for subliminal perception occur rarely. But unattended stimuli occur constantly, and they have the same influence on the Inferential System without become conscious. Subliminal presentations are just interesting evidence for the Inferential System. If you thought that anything in the brain could become conscious -- via say the spotlight of attention -- then subliminal presentations would be impossible. Either something was presented long enough to be perceived by the brain or it was not. Because the Inferential System receives all information and consciousness does not, it is perfectly reasonable that something could be found within the Inferential System, yet not be available to conscious introspection.

Conclusion

The Inferential System isn't just the product of my imagination, and I didn't construct it just to explain intuition. This essay first showed that the Inferential System produces one coherent conclusion. If anything, the newsworthy aspect of this is that the Inferential System itself apparently does contain the contradictory information.

We also had more about the automatic processing in the Inferential System, in that it also uses cues automatically. This includes cues that we are not attending to, and it includes cues that perhaps we could not be conscious of even if we were attending to them.