From Aaron Sloman Sun Jul  9 12:40:16 BST 2000
Subject: Re:      Query: Emotion and attention.
To: The ISRE List


Ronald de Sousa wrote:

[RS]
> Christine Tappolet of Montreal (whose excellent book Emotions et
> Valeurs is coming out very soon from Presses Universitaires de France)
> has asked me:
[CT]
>         "has there been any empirical research done
>         to test your claim that emotions direct our attention?
>         I like this claim, but I would like more scientific
>         reasons to adopt it! In general, I would be interested
>         to know whether there are philosophical or
>         psychological discussions of this thesis."

I wonder whether this is an empirical question or partly a conceptual
one.

Compare a different question: does intense pain direct attention?

Consider someone who claimed that it was possible to be in very intense
pain without paying attention to it, even when there's nothing else to
divert attention elsewhere, e.g. a life-threatening attack.

We might wonder whether such a person had learnt the ordinary concept of
"pain" and was perhaps confusing it with "injury". We know that injury,
even serious injury sometimes does not direct attention, e.g. under
anaesthetic.

But a state is not pain if there's not even the slightest disposition to
attract attention.

What "attracting/diverting attention" means is another question: it can
include behavioural events like switching direction of gaze, or changing
one's orientation in order to hear something. More interestingly, for
cognitively sophisticated organisms like us, it could refer to diverting
internal processing resources, and turning on or off certain forms of
processing, e.g. trying hard to recollect, dwelling on memories, trying
to predict, trying to decide, planning a route, trying to explain,
wondering whether ..., wondering why ..., etc..

These changes in processing need not involve any externally detectable
behavioural differences.

[RS]
> I myself confess that, with typical philosophical irresponsibility,
> I simply took this thesis (in my 1987 book) to be obvious.
> But surely some of the empirical researchers among us must
> have some relevant data?

Empirical data might be irrelevant if it's a conceptual question.

I don't think all the things people call emotional states are
necessarily connected with diversion of attention. However, in part this
is an unanswerable question because the word "emotion" is used in such a
loose and diverse variety of ways, both in ordinary discourse and in
scientific and philosophical discussion. Everyone relates it to their
own pet theory of emotions. (It's nearly as bad as "consciousness",
where sloppy terminology is the norm!)

E.g. some people think that love is an emotion. However, you can love
your family without being at all emotional about it. In fact most of the
time your love for your family is out of your mind: you are
concentrating on all sorts of other things, if you are a normal person,
not obsessive about your family.

Nevertheless, certain new percepts, or reports from others, about the
state of a member of your family are capable of interacting with that
dormant love to produce an emotional state, of concern, terror, joy,
pride, indignation, or whatever, depending on what you have just learnt
about about which member of your family.

Other kinds of dormant love can interact with new information to produce
different emotional states, e.g. jealousy, sexual arousal, a thrill of
joy (e.g. on seeing a smile of acknowledgement) etc.

But the other kinds of love can also be dormant for periods when other
things are the centre of attention (e.g. your work, or some episode
concerning a member of your family!)

So if "emotion" is *defined* to include such things as long term love
then it doesn't necessarily direct attention except in combination with
other things (like Nico Frijda's "concerns", which can be dormant for
a while.)

If however, you say that the enduring long term love is NOT an emotion
but an attitude (as some people do, including me), then that leaves open
the question whether the other things that are called emotions involve
redirection of attention, or at least a *tendency* to direct attention.
                          ^^^^^^^^^^^^^^^^^^^^^^^^

That underlined qualification (i.e. there's a disposition or tendency
which may not be realised) is required since there are so many competing
states that can redirect attention. Which one wins at any moment can be
a matter of changing dynamics and relative strengths.

E.g. a bright flash, or a telephone ringing, can redirect attention
without producing any emotion: it's just an automatic response.

The emotional states (not just attitudes) that are of most interest to
gossips, poets, playwrights, novelists, friends, etc. are the ones that
are hardest to investigate in laboratory conditions because they involve
semantically rich cognitive processing which need not be manifested in
behaviour.

Available behavioural repertoires may not even have sufficient
bandwidth! E.g. it may take a whole page of a novel to describe a
rich emotional state that rises and falls in a second or two.

If a certain sort of emotional state (the things we used to call a
"perturbance" here in Birmingham, and now call a "tertiary emotion") is
*defined* in terms of partial loss of control of attention, then it's
not an empirical question whether emotions direct attention. It's a
matter of how those emotions are defined.

Such emotions seem to be commonplace in real life (and consequently
also in novels, plays, etc.).

How often such states occur is an empirical question. Grief is an
example.

E.g. if you talk to people who are grieving, it is not uncommon to
discover that that state involves not being able to attend fully to all
sorts of tasks that the individual desires or even decides to attend to.
Instead their thoughts are constantly involuntarily redirected to
topics related to their recent loss.

If a person claimed to be in deep grief but also said that they were in
total control of their thoughts and could easily pay full attention to
any topic they chose to, then you might regard this as evidence against
the thesis if you thought the link was empirical.

However if it is a conceptual link you would suspect that "grief" was
the wrong word, and that it was not so much an emotional state as an
attitude of regret (i.e. a strong preference for the dead person still
to be alive: and preferences, even strong ones, need not be at all
emotional -- they are more like love for one's family).

Likewise if someone is deeply humiliated, infatuated with a member of
the opposite sex, obsessed with jealousy, full of excited anticipation,
etc. then, according to the common interpretations of those phrases,
that person will not find it easy to keep the objects of the emotion out
of his mind.

(E.g. look up dictionary definitions of "obsess", "obsession".)

For types of emotions which are conceptually linked to partial loss of
control of attention it would make no sense to do empirical research to
find out whether those emotions can redirect attention (compare doing
empirical research to find out whether all triangles have three sides).

Of course it would be appropriate to do empirical research to find out
under what conditions such states arise, how they affect attention, how
they interact with other emotional states, attitudes, beliefs, percepts.
etc, with which they are not conceptually linked, and also how far they
are culture-dependent, what their neural infrastructure is, etc. etc.

This topic is discussed in a lot more detail in this paper

    I.P. Wright, A. Sloman and L.P. Beaudoin,
        Towards a Design-Based Analysis of Emotional Episodes,
        Philosophy Psychiatry and Psychology, 1996, 3, 2, pp. 101--126

It is available in in the Cognition and Affect project directory
    http://www.cs.bham.ac.uk/research/cogaff/
along with many older and newer papers on this topic.

My guess (based on reading various books and journal articles on
emotions) is that much of the empirical research that is done is not
based on a good theory of the underlying information processing
architecture, which would determine the variety of metnal states that
are possible.

Several of the papers in our project directory are concerned to show
that if we start from an analysis of the information processing
architecture of human minds we can use that to define families of
concepts that are more or less closely related to our pre-theoretical
commonsense concepts.

Assuming different architectures (e.g. for newborn infants, or for other
animals), we find different families of applicable concepts.

In particular it turns out that there are at least three types of
phenomena that have been labelled "emotions". Following Damasio and
others we call the first two *primary* and *secondary* emotions. These
arise out evolutionarily relatively old portions of the architecture
(the reactive components, and the newer deliberative components
supporting `what if' reasoning capabilities - a common source of
secondary emotions).

A further class, *tertiary* emotions, involve newer, rarer, higher order
self-monitoring, self-evaluating, mechanisms that evolved much later
than the mechanisms required for the first two classes.

I don't know whether any other animals are capable of having tertiary
emotions. Perhaps chimps and bonobos are? Newborn human infants probably
can't because they don't yet have the third architectural layer
developed.

For more on this architecture-based approach to mental concepts see

    A. Sloman, Architecture-based conceptions of mind, in
        eds Peter Gardenfors, Katarzyna Kijania-Placek and
           Jan Wolenski, Proceedings 11th International Congress of
            Logic, Methodology and Philosophy of Science,
            (Synthese Library Series), Kluwer, to appear
        A preprint version is available in the cogaff directory.

As we understand the possible types of architectures better, we'll
better understand the variety of phenomena they can generate, and this
will provide an enriched ontology of mind,.

That's what happened when a good theory of the architecture of matter
led to an enriched ontology of matter (e.g. including explaining
isotopes, predicting new elements and chemical compounds, etc.)

Our work is much influenced by Herbert Simon's old but still
important paper:
    H. A. Simon, Motivational and emotional controls of cognition, 1967,
    Reprinted in  Models of Thought, Yale University Press, 29--38, 1979
(I think an earlier version was first published around 1962.)

Work by Johnson-Laird and Oatley on global interrupt signals is also
relevant:
    Oatley, K., Johnson-Laird, P.N.: `Towards a cognitive theory of
    emotions'  {\em Cognition and Emotion}, vol 1, 1987 pp 29--50

Returning to this:
[RS]
> I myself confess that, with typical philosophical irresponsibility,
> I simply took this thesis (in my 1987 book) to be obvious.

Conceptual truths are, in some sense, obvious, though people don't
always notice the obvious, especially if they have been trained only to
pay attention to measurements and statistical correlations.

Perhaps we can give them emotional jolts that will make them look
elsewhere?

Cheers.

Aaron
[Apologies for length: I felt detail was necessary in case there are
readers unfamiliar with the distinctions.]
===
Aaron Sloman, ( http://www.cs.bham.ac.uk/~axs/ )
School of Computer Science, The University of Birmingham, B15 2TT, UK
EMAIL   A.Sloman@cs.bham.ac.uk
PAPERS: http://www.cs.bham.ac.uk/research/cogaff/
TOOLS: http://www.cs.bham.ac.uk/research/poplog/freepoplog.html
Phone: +44-121-414-4775        Fax:   +44-121-414-4281/2799