[Heartlogic-dev] putting it all together (was Re: parameterizing)

[Joshua N Pritikin]

On Sun, Feb 08, 2004 at 05:57:21PM -0600, William L. Jarrold wrote:
> On Sun, 8 Feb 2004, Joshua N Pritikin wrote:
> > > Maybe you can help me by generating enough examples to motivate all three
> > > of APPRAISING-AGENT, OPINER, and POINT-OF-VIEW.
> >
> > OK, I'm going to review the parameters we have identified so far.  I
> > am also attaching a longer document which shows most of our detailed
> > discussion up to this point.
> >
> > PARTICIPATING-AGENTS: Associated with each appraisal are two
> > intentional agents.
> 
> Not exactly.  I think of participatingAgents as a binary predicate
> that maps between an event and an agent that participates in that
> event....Some events require at least two participant agents.  E.g.
> giving requires a giver and a receiver.  But
> AffectivelyAppraisingAnEvent requires only one agent, not two.

Yes, of course.  I have updated the working document.

> > APPRAISING-AGENT: Who is appraising the situation?  An appraisal is
> > the appraising-agent's opinion about or construal of the situation
> > cue.
> 
> Pretty much.  I'd prefer that the second agent read "An appraisal is the
> appraisaing agents affective construal of some situation."

Changed.

> > My proposal for adding one more parameter is to _split_
> > APPRAISING-AGENT into two parameters: OPINER and POINT-OF-VIEW.
> 
> Hmm.  Okay, I've read your note a few times.  It feels clearer.
> However, I'm embarrased to say that I strongly desire a name change.

No problem.

Needless to say, at this point I don't care about the terminology as
long as I can represent what I want to represent.

> It seems to me that what we want to do is keep #$appraisingAgent and
> #$mindreader and add a new binary predicate, #$metaMindReader.

That works for me.  I am also happy to follow your alternate proposal
at the end of this email.

> > Parity check: The MINDREADER parameter is unchanged.
> >
> >   If the appraisal is #3 and MINDREADER = KM then we still ask,
> >   "Mr. Computer Model, how does Toby feel about receiving Hot Coco from
> >   Daddy?"
> 
> For Case #3 I'd say...
> 
> #$appraisingAgent = Toby
> #$mindreader = Toby
> #$metaMindReader = KM
> 
> >   If the appraisal is #4 and MINDREADER = WLJ then we still ask, "WLJ,
> >   how does Toby imagine Daddy feels about giving Hot Coco to Toby?"
> 
> The immediately above would be represented like so...
> 
> #$appraisingAgent = Daddy
> #$mindreader = Toby
> #$metaMindreader = WLJ

This seems to convince me that we are talking about the same thing.

> > Hopefully you have grasped the proposal now.  If you prefer a
> > different terminology than OPINER / POINT-OF-VIEW then this is a good
> > time to change.
> 
> So, I guess I understand where you are wanting to go and I have
> made my point about a name change.
> 
> I'll add this.  Mindreading is an event too, just like appraising it.
> Sometimes the object of a mindread is an appraisal. Other times, the
> object of the mindread is a mindread of an appraisal.  Thus, if we go
> this route the predicate #$metaMindReader will be superfluous.  To refer
> to a meta-mindreader, we need just create a mindread of a mindread event.

This proposal is also fine.

> I could spell this out if you'd like.  The predicate #$metaMindReader will
> paint us in to a corner.  For example, if we have two people mind reading
> about "How does Toby imagine Daddy feels about giving Hot Coco to
> Toby?" then we'd be better off by having two events for each of their
> meta-mindreads rather than one slotized thing.  WHy?  Well, if we go
> with one slotized thing, i.e. metaMindreader, then how will we
> correctly associate the two resultant distinct metamindreads with the
> appropriate metamindreaders.  I've skipped over lotsa stuff here and
> thus have not really parity checked.  Maybe I'll go slower in the next
> round.

No, I got it.

> So, in sum, I think I just want slots that map like so...
> 
> #$participatingAgents: Event X Agent
> 
> #$appraisingAgent: AppraisalEvent X Agent
> 
> #$mindreader: MindreadingEvent X Agent
> 
> #$giver: GivingEvent X Agent.
> #$receiver: GivingEvent X Agent.

Yah.

> We'll need lots of other slots to connect MindreadingEvents to
> AppraisalEvents and so on.
> 
> Then again, maybe I am missing something about where you are wanting to go
> with this.

No, that sounds fine.

I believe we have discussed all the concepts needed for this next
part:  Let me try to propose a type of hypothesis which we can
explore, building on the ideas from your dissertation.

Given this situation:

  Goal: Tracy wants a banana.
  Situation: Mommy gives Tracy an apple for lunch.

  #$participatingAgents = Mommy, Tracy

And given two appraisals in the same context except for the
#$appraisingAgent:

1.
  #$mindreader = Tracy
  #$appraisingAgent = Tracy
  English = How does Tracy feel about receiving an apple for lunch?
  Appraisal = Tracy wanted a banana.  Because Tracy is inflexible
              and bratty, any fruit besides a banana fulfills
              one of her AntiGoals.
  Tracy's goal status = AntiGoal

2.
  #$mindreader = Tracy
  #$appraisingAgent = Mommy
  English = How does Tracy imagine Mommy feels about giving
            her an apple for lunch?
  Appraisal = Mommy knew Tracy only wanted a banana.  Because Mommy is
              cruel and nasty, Mommy deliberately gave
              Tracy something she didn't want.
  Mommy's goal status = AntiGoal

Hypothesis:

If both appraisals have a goal status of #$AntiGoal then the overall
situation fits the general category of #$Protest.

(The #$metaMindReader or second-order #$mindReader works as usual.  KM
or WLJ or Joe Blow will be determining or rating the output of this
hypothesis.)

+ + +

In other words, when there exist two appraisals of the same situation
by the same #$mindreader and differing #$appraisingAgent (as above)
then there is a hypothesis about how to "combine" the two appraisal,
primarily according to their goal status.

I predict that this hypothesis can give precise, formalized meaning to
#$Protest and at least 10 other affective words.  (Of course, this has
not been verified empirically..)

Am I going too fast?

-- 
A new cognitive theory of emotion, http://openheartlogic.org