EECS 690

1
EECS 690

• April 25/27

2
The relationships between morality and emotions

• Certainly, emotions like shame or guilt are
  strongly connected to moral situations and may
  motivate moral behavior.
• However, claims that morality just is a set of
  emotional reactions is a violation of the
  is/ought distinction.
• Certainly emotions and feelings provide us with
  important information quickly, and considerations
  of the feelings and emotions of others is an
  ethically relevant concern.

3
Feelings/emotions distinction

• A feeling involves something visceral. A feeling
  of pain is not itself and emotion, but may cause
  emotions.
• Emotions often involve some feelings or other,
  but the emotions isnt defined in terms of just
  the feelings.
• Feelings and Emotions are generally lumped into a
  category called by some Affective States (so
  called because they are things that just happen
  to you, i.e. things that affect you, not things
  you effect) and called Conative States by
  others (as opposed to Cognitive States).

4
Emotional states as information

• Facial expression, non-verbal vocal utterances,
  body language, etc. are all ways in which people
  communicate vast amounts of information to each
  other, so any system that was able to (even in a
  limited way) read these cues, could interact much
  more efficiently with human beings.
• An application to assist sufferers of Aspergers
  syndrome, many of whom have significant
  difficulty interpreting facial expressions. Such
  an application could be a prosthetic for social
  interaction the way that a prosthetic leg assist
  in walking or the way a notepad becomes a
  prosthetic memory.

5
Affective/conative communication

• Breaking up facial expressions, body movements,
  postures, vocal tone, etc. into their discrete
  parts to make use of such information for the
  purposes of communication is one thing that
  computers are getting more and more able to do.
  It is easy to see why this might be important,
  but does a system need affective or conative
  states of its own to make moral decisions?

6
Three general viewpoints

• Mainstream view (dates back to even pre-Socratic
  philosophy) Moral reasoning and decision-making
  should be devoid of emotional bias.
• Humes view (others have subscribed to this view,
  but Hume is the best early proponent of such a
  view) Conative States (a term invented by Hume)
  like sympathy and empathy form the backbone of
  the ethical conventions of civil society.
• Aristotle Affective states influence the way we
  behave, and so they are good insofar as they
  encourage virtuous behavior, and should be held
  in check insofar as they do not.

7
Five categories of emotion theory

• This breakdown is owed to Jesse Prinz
• Product theories
• Feeling Theories emphasizes the conscious
  experience of emotional states
• Behavior Theories identify emotions with
  specific behavioral responses
• Process theories
• Somatic Theories emphasize the bodily processes
  associated with emotions
• Processing-mode Theories emphasize the role of
  emotions in modulating other mental activities
• Cognitive Theories emphasize the role of beliefs
  in emotions

8
From Buridans Ass to Emotional Heuristics.

• Medieval philosopher Jean Buridan described a
  thought experiment in which an ass starves to
  death equidistant between two equivalent bales of
  hay.
• The lesson of Buridans Ass is that motivation
  and decision has to be at least partly
  affective/conative. Contemporary thinkers point
  to emotional heuristics as reinforced
  short-cuts for performing complex tasks. Much of
  this work was pioneered by Herbert Simon, who won
  a Nobel Prize in Economics in 1978 for his work.

9
Functionalism and Somatic Theories

• A robot determined to allow no harm to a human
  being would need to be aware of the facts of
  human pain sensitivity. A quick way to this
  knowledge is to have the same sensitivities.
  (p.152)
• Somatic theories strive to produce the functional
  equivalents of certain affective/conative states.
  Such concerns as above indicate the importance
  or potential usefulness for systems with such
  capacities, but doesnt answer the question of
  whether a system needs to have its own
  affective/conative states to be moral.