• Mental models are more than maps; they're causal explanations of system behaviour.
• Schemata are organisational structures that represent knowledge.
• Pure schema theorists believe procedural knowledge is also part of schemata.

Date: Fri, 5 Jul 1996 10:43:08 +1100
From: Clark N Quinn 
To: resodlaa@usq.edu.au
Subject: resodlaa: question - schemata/mental models - and procedures

Dear Lorraine,

>Need clarification, please.

One answer....and I highly recommend the following reference to sort this
out (that's who I learned it from):

  Rumelhart, D. E., & Norman, D. A. (1988).  Representation in memory.  In
R. C. Atkinson, J. J. Herrnstein, G. Lindzey, & R. D. Luce (Eds.) Handbook
of experimental psychology.  New York: Wiley.

>Mental models - I see these as Kintsch does - sort of as mental maps.

I'll disagree.  Mental maps or images are static.  I see mental models as
dynamic.  That is, they're causal explanations of system behaviour.  They
are used to explain observed outcomes and to predict the effects of
perturbations.  I believe that there hasn't been enough effort to help kids
acquire the ability to reason causally, but some new tools to support
qualitative modelling, such as Apple's KidSim, as well as the existing
quantitative tools like Stella, help bridge this.

>Schemata - these are cognitive structures containing word lists, images,
>and propositions (from Gagne & Yekovich squared's text).

This is a limited view of schemas.  Schemas are an organisational structure
to represent knowledge.  They were proposed to explain inheritance and
defaults which were difficult for semantic networks to easily handle.  They
sort of arose simulataneously as Minsky's 'frames', Rumelhart's 'schemas',
and Shank's 'scripts'.  However, knowledge is more than just declarative.

>Gelman & Greeno talk about "action schemata" for kids who are trying to
>gain competence in the area of set theory and math.  To me, anything
>involving an action is going to involve a procedure, but a procedure
>isn't a word list, image, or proposition.  So My "book learning"
>definitions seem to be too narrow.

For declarative information they can be the static structures considered
above.  But they also include the proceduralisation of applying knowledge
to particular problems.  So, you may have been presented with a declarative
definition of the process of instructional design, but as you perform the
task you gradually acquire procedural knowledge that pure schema theorists
believe are also part of the schema.

>Now, where, exactly, do procedures fit in?  There seems to be a difference
>of opinion and "fuzziness" about this, and we've had long e-mail
>discussions in our doc seminar about this issue.  I also spoke to a
>cognitive psychologist at Ed-Media, who said the action schemata fit into
>mental models.  So now I am confused!

No wonder you're confused.  It may help to realise that schemas are a
'fuzzy' concept.  Schemas are attributed with rich and powerful properties
but trying to be concrete about them is like the proverbial 'nailing jelly
to a tree'.  Note that Rumelhart found it difficult to determine how such
wonderfully flexible behaviour as schemas could be implemented in the
formal models being touted, which eventually led to the Parallel
Distributed Processing books and linking with other connectionist
approaches.  Anderson separates static declarative knowledge from
procedural knowledge in his ACT* representation.  However, while they may
not be a full explanation of cognition, at a high level (e.g. for
describing some aspects of learning) they are a useful explanatory tool.

As for the EdMedia explanation, I would put it the other way around, that
mental models are action schemata they've got organisation, that is the
part/whole relationships, and the causal links).  However, you could argue
that schemata for certain low-level causal actions, such as hydraulic
pressure, could fit into a mental model of a car jack.  That's just an
example of the 'embedding' property of schemas, I'd argue, where your
schema for wheel embeds into your schema for automobile.

There are other opinions of what mental models are, including Phil
Johnson-Laird's view that all thinking is done by mental models.  This is
too much for me, and I prefer the conceptualisation described above.

>I'd very much appreciate it if some
>expert could enlighten me and make the distinction clear.

I won't claim to be an expert, but this is what I have learned.  I hope
this helps, and I'm happy to negotiate understanding further.  -- Clark



---------------------------------------------------------------------
 _--_|\    Dr Clark N. Quinn                         Open Net Pty Ltd
/      \   Multimedia Manager                     210 Clarence Street
\_.--._*   P: +61-2-9267-1222 x103                    Sydney NSW 2000
      v    F: +61-2-9373-2703                               AUSTRALIA
           cquinn@opennet.net.au           http://www.opennet.net.au/
---------------------------------------------------------------------