The Emotion Machine

We’ve already seen how subgoals can be connected to serve our goals—the way that Use Spoon could become attached to Fill Cup. But when your imprimer praises you, some machinery elevates your current goal, to make it more ‘respectable’ by raising its place in your cloud of goals.

However, this image tells us nothing about how those processes actually work, so we need to construct some theories about how attachment works to ‘elevate’ goals. First, this must depend on circuits that recognize when the praise comes from an imprimer:

Student: Why did you insist that those “AND” devices should require both praise and an Imprimer?

That’s because, as we noted in §2-3, we would all be in danger if praise, alone, could cause our brains to elevate goals—because then any stranger could program us by suggesting new goals and then praising us.

Student: But to some extent that’s already true; I am not immune to compliments—even from persons I don’t respect.

One feature of human diversity is that we can learn the same things in different ways—and any psychological event is likely to have several causes. If attachment-based learning exists, it is only one part of the story.

Student: But something is missing from this scheme because, even after its level is raised, that ‘fill cup” goal is still floating around with no connections that could get it aroused.

Indeed this idea is incomplete. There is no use to learning something new unless one also has ways to retrieve it when it is relevant. This raises many questions like these:

To what should each new goal be attached?

When and how should it be aroused?

What kind of priority should it have?

How long to pursue it, before giving up?

There are no simple answers to these, because all those issues must involve much of the rest of our mental machinery. Nevertheless, it is hard to see how to think about such things without a set of ideas about ‘levels’ of mental activities. Our brains have many systems that learn—and as these develop over the years, they may tend to form roughly hierarchical structures, because each fragment of newly acquired knowledge is built upon things that we’ve learned before.

For example, in the course of everyday thinking, you need to constantly control the “level of detail” of descriptions. When a plan seems to be working successfully, you’ll want to “descend” to work out details—but when you seem to be getting stuck, you’ll want to ‘look up’ to a higher-level overview, instead of investing time on subgoals that may not be relevant. [See §§Level-Bands]

???????????????????

§2-5. Learning and pleasure

When Carol was trying to fill her pail, she had to try several experiments before she succeeded by using her spoon. When she recognized that her goal was achieved, she felt satisfaction and a sense of reward—and then those pleasant feelings somehow helped her to learn and remember. So this process involved a good many steps:

Carol filled the pail with her spoon.

She recognized that her goal was achieved.

Then she felt pleased with her success.

Then, somehow, that pleasure helped her to remember.

Now we’re glad that she felt gratified—but what functions did all those feelings serve, and why should that process take so many steps? What sort of role might pleasure play in how we construct our memories? Why couldn’t Carol just simply remember which methods worked and which ones failed?

The answer is that ‘remembering’ is not simple at all. On the surface, it might seem easy enough—like dropping a note into a box, and then taking it out when you need it. But when we look more closely at this, we see that it involves a good many steps: You first must select which items that note should contain, and find adequate ways to represent them—and then you must give them some set of connections, so that after you store those parts away, you’ll be able to reassemble them.

Citizen: Some say that our brains remember everything so, that if you cannot recall some event, some part of your brain must be suppressing it.

This ‘photographic memory’ myth is not supported by evidence; the consensus from many experiments is that we don’t remember nearly so much. [See §6-2]

Student: What about the old idea that, for each of our accomplishments, we just ‘reinforce’ our successful reactions? In other words, we simply connect the problem we faced to the actions that led to our solving it.

That is a simplistic way to describe how learning might work, when seen from outside—but it doesn’t explain what might happen inside. For, neither ‘the problem we faced’ nor ‘the actions we took’ are simple units that we can connect—so, first you must choose some way to describe both the ‘If’ and the ‘Then’ of that pair of events. Then, the quality of what you learn will depend on the natures of both those descriptions.

Thus, for Carol to learn, her brain must construct some descriptions of which methods worked—as well as, perhaps, of which methods failed. But after her struggle to fill her cup, which of all the things she did should get credit for her final success? Should Carol attribute her success to which pair of shoes she was wearing then, or the place in which that event occurred, or whether the weather was cloudy or clear? What if she smiled while using that fork, but happened to frown when using that spoon; what keeps her from learning irrelevant rules like, “To fill a cup, it helps to frown?”

In other words, when humans learn, it is not just a matter of making connections but of constructing the structures that those connections connect—and no theory of learning can be complete unless it also accounts for this. Furthermore, we may need to represent not only those external events, but also some relevant mental events. Thus Carol will need some machinery to decide which of the thoughts she was thinking then should be represented in what she remembers. And she will need