re: double-loop learning
Fri, 1 Nov 1996 23:47:10 +0100

Hi Magnus,

Good to have such a speedy response to my drivellings. At one point you say,

>welll, as I understand the concept, one of its most iteresting features it
>that it implies the _existence_ of a second loop, in situations and in
>relations where we act as if only a single loop exists. The first point, as
>I see it, is not to choose among different functions (as understand as
>"strategies" in Argyris terms, quite commonsenical), but to focus attention
>to and become aware of the _existing_ strategy. It is about beginning to
>_see_ the system, and thus learn about it.

Yes. This capacity for becoming aware of the single loop is what's implied
in Pask's Finite Function Machine idea. Clearly, any superordinate language
is capable of comprehending discourses in its subordinate languages.

But the less obvious thing asserted by Pask's notion, which maps on in an
interesting way to your account of Argyris' notion of "strategy", is that a
lower-level language is _incapable_ of comprehending itself fully: you
_have_ to have a higher-level language for that, i.e. to "focus attention
to and become aware of the -existing_ strategy", as you put it. (This is
asserted explicitly in Godel's "Incompleteness Theorem", which I mentioned
in my last e-mail.)

Another way of saying this is that it's impossible to make full sense of
your circumstances without some way of "stepping outside" your

Actually, pcp has two concepts which address this issue.

First is the notion of a construct hierarchy, with superordinate and
subordinate relationships, in which the highest-level constructs, core
constructs, are identified with personal values: which is what I was
banging on about last time.

But the second notion is expressed in Kelly's Sociality Corollary, if you
think about it: one way of "stepping outside" one's own circumstances is to
try and view them through _another person's_ constructs: doing an exchange
grid, and so on.

You go on to say:

>I see that the idea of core constructs, and costructs organized
>hierarchically, is similar to Argyris' view. What I do not see here, or have
>any clear understandig about, is the idea about theory-in-use and espoused
>theory. One of the main points with double loop learning is to discover the
>strategies and values I actually act upon, regardless of whether they are my
>espoused values or not. The point becomes to become aware about
>inconsistencies in my construct structure, as I see it, between espoused
>constructs and enacted constructs. But perhaps that point is too trivial

No, not at all. Actually, the Finite Function Machine isn't capable of
embodying the distinction between strategies and values actually acted
upon, and strategies and values which are "merely" espoused. Sure, you
could define a lower-level function (f) which stands for the espoused, and
another function (g) which stands for the acted upon, and see the
higher-level part of the system as always selecting (g) regardless of (f):
but this is just playing about with the model; the model's not designed to
represent the important distinction you're highlighting.

Actually, you've raised a very interesting issue here, and I'd appreciate a
reference to any of Argyris' "double-loop" work which tackles the
espoused-enacted distinction you've highlighted.

Recently, I've become interested in the measurement of personal values
using resistance-to-change techniques, and have had to confront the problem
that in many empirical investigations, there's a real problem with the
"social desirability effect": viz., that people may tell you what they feel
you anticipate they should believe, rather than what they actually believe
and which guides their actions.

There's a rather neat way of trying to handle this when doing grid-based
studies of the personal values of a group of respondents (Jankowicz, 1996,
shout if you want a copy). Basically, you split the repertoire of values
mentioned by the group into two sets for each person: those to which the
person subscribes in action; and those of which the person's aware and
espouses (because they're values and hence, by definition, "valued" by
anyone who shares in that groups' culture), but which s/he _doesn't_
necessarily utilise to guide action.

So: it would be very handy to hear Argyris' views on this, and any
reference that you feel would illuminate this issue from his point of view.

Finally, re.;
>I become a bit unsure about how you see the learning process in the person.
>For all the benefit of the metafor with the finite functionmachine, I
>believe it hides important ascpects of the learning process Argyris is
>focusing. One of the main points in his later works is that we act in ways
>that inhibit our learning.

Sure. As I mention above, the Finite Function Machine isn't a model
designed to incorporate this last point in its structure: i.e., it doesn't
specify a mechanism for predicting when a person will act in ways which
inhibit learning, and when s/he won't. There again, nor does Kelly's notion
of a construct system. In both cases, you can _use_ the system to talk
about acting in ways which might inhibit learning, but there's nothing
about the model which represents an _inbuilt_ prediction about which will

In pcp, you'd simply have to ask "in what way does it seem functional for
the person to act in a way which s/he doesn't espouse; what to him/her, is
the function served by acting in a way which inhibits learning?" And you'd
move the analysis onwards by pointing out that the distinction between acts
which inhibit learning and acts which don't, implies the existence of an
external person who has the privilege of recognising when a person is
acting effectively and when s/he is acting "despite him/her-self", i.e. in
a way which "inhibits learning".

In Pask's approach, you could model this by means of two Finite Function
Machines coupled to interact on each others' functions.

But ultimately, in pcp, you acknowledge that the final arbiter is the
person him- or her-self: a person always acts in ways which serve his or
her objectives _as s/he construes them_; but a person "acts in ways which
inhibit his/her learning" when the acts seem to the outside observer not to
be congruent with that person's objectives-- as the external observer
construes the person's construing of them!

When, additionally, the person _him/her-self_ acknowledges that the
actions seem incongruent with objectives, we talk of situations calling for
therapy: admission that something's not quite as it should be, the "cry for
help", etc. (As I understand this. But I'm not a clinician, and maybe any
pcp clinicals reading this could chip in and correct me where I'm

Significantly, Pask's model doesn't lend itself too well to this latter

>Have a good weekend!

Ta. You too!

Kindest regards,

Devi Jankowicz.

Jankowicz, A.D. (1996) "Personal values among public sector employees: a
methodological study". A paper given at the 3rd. Conference of the European
Personal Construct Association, University of Reading, UK., April.