Path: utzoo!attcan!utgpu!jarvis.csri.toronto.edu!rutgers!tut.cis.ohio-state.edu!gem.mps.ohio-state.edu!ctrsol!srcsip!orion!mccarthy
From: mccarthy@SRC.Honeywell.COM (Jim McCarthy)
Newsgroups: comp.cog-eng
Subject: Re: What is the difference between a metaphor and a mental model?
Message-ID: <36145@srcsip.UUCP>
Date: 24 Oct 89 15:09:18 GMT
References: <31966@ucbvax.BERKELEY.EDU> <1215@kuling.UUCP>
Sender: news@src.honeywell.COM
Lines: 50
In-reply-to: larso@emil.CSD.UU.Se's message of 23 Oct 89 14:49:25 GMT



In his posting of 10/23/89, Lars Oestreicher stated:

> The form of the mental model, i.e., it's "shape in the mind" is in most
> cases impossible to define or study.  In my opinion the representation
> issue constitutes a less interesting question.

My vested interests (I'm doing research on this topic) prevents me from
agreeing.

At the annual meeting of the Human Factors Society last year, and again
this year, my colleague (and fellow graduate student) Mark Cushman and I
presented ongoing research posters detailing our work in knowledge
representation.  In brief, we discussed the relative efficacy (in obtaining
valid representations) of two data collection techniques (card sorting and
pair-wise comparisons) and three representational tools (Mutidimensional
scaling, Cluster analysis, and the PATHFINDER algorithm).  Again in brief,
we found that there were clear expert/novice differences in the
representations obtained (confirming earlier work by others), that, within
the Clustering solutions, there was a qualitative difference in the
representations stemming from the card sorting and paired comparison
techniques, but that, although the experts demonstrated a clear preference
for the card sorting solutions, these representations did not differ in
there validity (where validity was defined as the ability to complete
partial clusters).  That is, the card sorting derived clusters were
completed just as accurately as the paired comparison derived clusters.

My point is this, using techniques such as those discussed here (as well as
others) researchers are able to obtain functionally useful knowledge
representations.  I believe that these representation give us a peek at the
mental models resident in experts and novices in a number of domains.

Are these representations useful?  I believe they are.  Mark and I are just
beginning a new cycle of research aimed at determing the usefulness of
these representations in instructional (as "advanced organizers" or
"anticipatory sets") and evaluative settings.  Further, it seems to me that
such representations could greatly aid in the way information is presented.
For example, I can imagine an intelligent menu system which gradually
changes its structure as a novice user becomes more expert-like in his/her
mental model of the software's function.

Because this research line is something of a hobby for Mark and I as we
continue our graduate careers is other areas, it is possible we are taking
a nieve look at the literature.  I'm interested if others on the net feel
as we do that these representations portray mental models, and that they
are potentially useful in terms of instruction, evaluation, etc..  If you
don't feel like posting your opinions please feel free to email me.

Jim McCarthy