Dear Terry,

Thanks for your contributions to the theory thread on PhD-Design. I feel
that my understanding of the subtleties of the discussion is not what it
should be, and so I have a number of questions for you from your first
post that I still haven't been able to get my head around. I hope you
don't mind me chasing you up. If you have the time to respond, I'd be
grateful.

When you say that the causal explanation "of agency and behaviour lie in
understanding the physicality of how people function", I'm not sure I
understand you. It appears to me that you are advocating a (purely)
materialist explanation of human behaviour. Do I understand you
correctly?

In that case, any appeal to phenomenological categories (such as
thoughts, intentions, motivations, feelings etc.) cannot lead to any
causal explanation (as you then say); but (I think) what you imply is
that human agency can ultimately be explained (only) through physics
(which would make agency an illusion, since physical explanations are
deterministic). If that's the case, I don't then understand the role of
the phenomenological categories. While you acknowledge phenomenological
categories, you say they don't have the same ontological or
epistemological status, and cannot feature in causal explanations. Is
this a type of epiphenomenalism? Or would you say that these categories
more like obsolete linguistic conventions that we have inherited and
would be better off discarding (at least for epistemological reasons)?

Something else I don't understand: while it is surely true that
"literatures on phenomenology, critical theory, hermeneutics, grounded
theory, alternative paradigms, post-positivism, symbolic interactionism
etc can be seen as different ways of minimising ambiguity in theory", I
thought that underlying most of these was some fundamental disagreement
(among others) about the nature and role of theory in representing and
understanding (social) reality. Certainly this is the case with
phenomenology and (philosophical) hermeneutics, at least. Does this
square with your understanding? Aren't there deeper-level philosophical
disagreements as a matter of course with some of these other
perspectives that preclude them from being considered just one more
strut in some epistemological, superstructural ideal? I realise here
that I may be reading too much into your comment that these perspectives
are different ways of minimising ambiguity in theory, but this is where
Klaus' post(s) appear to have the same concern. Certain strands of
phenomenology and hermeneutics explicitly deal with some of Klaus'
points, such as creating/maintaining a role for human agency, and
dissolving the Cartesian, dispassionate-subject view of knowledge of the
external world.

I'd also like to query your usage of the term "theory". Again, depending
on how well (or not) I understand you, I think you use "theory" to be
equivalent to "causal explanation". I get this impression from your
comments concerning correlatory representations (you don't say
"correlatory theories"). In referring, for instance, to Schon's work as
constructing correlatory representation (rather than constructing
theory), perhaps I am gaining a glimpse of the radically different usage
of "theory" that you envision to be epistemologically useful. If Schon
did not construct a design theory, then was (even) Simon not engaged in
theorising about design? Is Grounded Theory then about correlation (and
not really about theory)? And is anything under the label of "Social
Theory" (e.g. Marx, Adorno, Mead, Vygotsky) actually "theory" by this
definition?

I am convinced of the value of careful, consistent and formal
definitions in the advancement of scholarship, and I think your papers
and posts have been vital resources for design researchers towards this
end. But in the field of design, which sits astride so many traditional
disciplinary boundaries, is it problematic to appropriate a term that
has such a rich and useful history of use in a discipline that feeds
into design research (such as social theory)? Would we be better off
differentiating the epistemologically valid "theory" that you describe
as "formal theory" or "causal theory" or "epistemologically valid
theory"?

I think these are vital (and epidemic!) issues for design research, and
things that I need to understand better. Thanks very much for your time
and scholarship.

Kind regards,

Ben Matthews


Terry Love wrote:

As far as I can see, the most significant split in theory is that
between causal and correlatory approaches. This is especially
problematic when correlatory models of relationships between 'inputs'
and 'outputs' are carelessly or accidentally regarded as causal
representations.  This problem can be found in much the same form in
systems disciplines and in those disciplines that focus on stimuli and
responses or causal factors and consequences. The depth and consequence
of this problem are great because it is so extensive. The problem stems
from epistemological category confusion due in part to limitations in
data gathering techniques. There are many disciplines in which theory is
made about why people act in certain ways in particular circumstances
(e.g. history, economics, psychology, management, marketing, and design
research). The causal explanation, of agency and behaviour, lie in
understanding the physicality of how people function. Non-physical
theories (e.g. thinking, emotion, motivation, group behaviour,
attitudes, personality attributes) are epistemologically in a different
frame that cannot provide direct causal representation. These other
approaches, however, have the advantage that correlatory data is
relatively readily accessible. The difference in convenience, however,
might mask but does not avoid the difficulty that, due to their
epistemological and ontological status, they cannot provide casual
explanation.

This causal-correlation problem of theory making about human functioning
has led to many confusions being introduced into the literatures of
theory making.  Stepping back a bit, and going to Popper's main work on
theory making,  'Unended Quest' (Popper, 1976), he suggests that theory,
external observations and subjective perceptions lie in three
incommensurate 'worlds'. Put simply, Popper argues that it is not
possible to prove theory as true or false by observations or subjective
perceptions. Instead, he argues that theory is only validated in terms
of other tightly defined theories. This is now widely accepted in the
appropriate fields of philosophy (the same also applies in relation to
the other incommensurabilities, i.e. its not possible to prove
subjective perception by external observation or theory, and its not
possible to prove objective observation by subjective perception or
theory). In many theory literatures is found extensive evidence that it
has been assumed that correlatory evidence creates theory.  This is a
significant issue because it confuses distinctions between, concept,
correlation, theory, model, prototype, problem definition,
representation, internal imagogenic perception, feeling, emotion and a
host of other concepts.

The causal- correlatory split in theory making differentiates types of
theory and ways of building theory. A causal representation of a
situation is a result of deep understanding of how a situation functions
in terms of its physicality. This may involve concepts that are simply
place markers (e.g. the use of the concept of 'glucose tolerance factor'
to represent something that is yet unknown in theory representations of
the physiology of hypoglycemia). Its primary focus is, however, to
create theory that as representations has strong mapping onto the
physical causal processes.

Correlation-based representation contrasts with causally based theory in
many ways . The development of  correlation-based representations is via
the identification of statistically significant correlations between
'causal factors' and 'outcomes' to fix the constants in a mathematical
formula (e) that emulates the data. There are an infinite number of
formulae that can be used for this task and there is no requirement that
the formula (e) chosen map in any way onto the causal mechanisms. In
fact, they are often chosen for administrative convenience. If the
phenomena is relatively smooth and unchanging in its behaviour then
correlatory formula offer the basis for interpolating between data, and
for projecting a small way beyond data. In situations that are marked by
strong changes (such as many design situations) correlatory
representation of data offers little utility beyond simplifying the
presentation of the data. It is questionable as to whether this
justifies them being viewed as 'theory'.

Entirely a different problem, and needing more time and space than I
have is the situation in which simplified system models are based on
correlatory data. These models are commonplace in the literatures of
management, systems, education and learning. Characteristic of these
models is a lack of epistemological consistency (now increasingly
recognized as a serious weakness of systems approaches to modelling); a
foundation in correlatory rather than casual understanding of phenomena;
and a mode of representation that sacrifices correspondence with
underlying theory for simplified conceptualisation. That is, there has
been a shift from a theory representation to a reformulation as tool for
teaching or managing. In terms of theory development, this raises
concerns if the underlying theory becomes accidentally (or carelessly)
reformulated such that the teaching tool is regarded as the theory (the
unfortunate status of the teaching models of Kolb, Schon and other
researchers working in management and education).

_________________________________
Ben Matthews
Information Environments Program
University of Queensland
Brisbane Australia 4072
Ph +61 7 3365 8308