Emotion in Reason (Was: Plea)

Eliezer Yudkowsky (sentience@pobox.com)
Sun, 19 Jan 1997 00:38:37 -0600

I recently had an interesting insight into the eternal conflict between
emotion and intellect: I finally realized how such a conflict is even
theoretically possible.

I once took the MMPI (Minnesota Multiphasic Personality Inventory).
Aside from my disagreement with the touchy-feely nature of the questions
and the astrological-oid personality assessments, my main issue was with
questions like: "Do you follow your head, or your heart?" I never
perceived the two as being in conflict, and I didn't see how they could

Just recently, I realized that what was meant by the question was a
conflict between two emotions, one of which had stronger rational
support. Okay, now I get it. The answer to the MMPI question: I don't

Why was I baffled earlier? Because I saw emotions and intelligence as
belonging to two different spectrums of the brain. In retrospect,
emotions are the affective aspects of our goal-oriented cognitive
systems while intelligence is, well, everything else. Reasoning,
maybe. How can a goal-object come into conflict with a
reasoning-object? Doesn't make sense. Ah, but reasoning-objects give
rise to goal-objects, so you can have one goal strongly supported by
reasoning but of low perceived value, and a contradictory goal of high
value supported by the ur-goal emotions - survival, eating, sleeping,

The basic conflict seems to derive from the two primary slots of goal
objects: Justification and value. It's possible to have goals of
complex, high justification but low value, or simple, elementary
justification but high value. Under such circumstances, I go with the
high-justification goal if I have adequate mental energy or
self-control, or branch a sub-goal of going with the high goal if I

An example of an "emotion-reason conflict" might be working vs.
sleeping. Working has many rational justifications, but the actual act
of work is not itself rewarding and resistance builds up. This
resistance, which acts along a behavioral rather than justificational
channel and which thus is not subject to simple mental control, reduces
the perceived value of work. Sleeping has no rational justification (if
adequate sleep has been acheived) but is an ur-goal, an emotional
primitive, and thus has a high perceived *value* even if fewer

This is of course only one class of "emotion-reason conflicts", that
dealing with clashes along the "value" slot. There are also clashes
along the "justification" slot, such as intuition, or experience,
conflicting with reason. From analogy and personal experience, giving
an example is left to the reader.

The generalized pattern in both cases seems to be an ur-value or
ur-justification coming into conflict with a rational justification
sequence. Ur-values are emotions; ur-justifications are generally
called "intuitions". In the case of emotions, I try to follow the
rational justification sequence because that's what works; in the case
of intuitions, experience shows that the best course is to incorporate
my intuitions in the opposing logic, and if that fails... I don't know;
it never does.

This concludes this week's lecture on cognitive science and "How To
Think." Comments?

         sentience@pobox.com      Eliezer S. Yudkowsky
Disclaimer:  Unless otherwise specified, I'm not telling you
everything I think I know.