Implicit Cognitive
Therapy
Readings
Notes from Meetings
- January 25, 2006
- Meeting attendees: Jamie, Kenny, Mark, Forrest, Martin
- Started with a discussion of how everyone in the group might
potentially be related to this research program. Kenny originally gave
a presentation to the cognitive area about cognitive therapy, which
highlighted the differences between the ways that clinical and
cognitive psychologists view cognition. One of the most notable
differences was the focus on unconscious/implicit cognition in
cognitive psychology which was lacking in clinical psychology. The idea
of writing a grant to try to use implicit tasks as part of therapy
surfaced again in Mark's class on unconscious cognition, which was
attended by Jamie, Joe and Chris. All four of these individuals
expressed an interest in persuing work on this topic. One suggestion of
Mark's was that computer-based therapy could be particularly useful for
rural populations since it would not require direct intervention of a
therapist. Forrest has previous experience in this area, and in a
conversation with Jamie indicated his interest in the project. For his
dissertation, Martin plans to administer a form of CBT over PDAs. It
might also be possible to perform implicit cognitive therapy in a
similar way, and so Martin was asked to join the group.
- Forrest mentioned that there currently is a body of work on
automatic thoughts in the clinical literature that might be worth
investigating.
- The basic focus of implicit cognitive therapy would be to
change either the accessibility of particular concepts, the
associations between particular concepts, or the attention that is
automatically given to particular types of stimuli.
- Forrest and Kenny mentioned that there is relatively little
work even investigating the use of implicit measures in relation to
therapy, and so we might be better off starting with assessment issues
instead of immediately trying to design interventions.
- Performance on an implicit measure might be particularly
related to how well someone behaves when they are in stressful or
distracting enviornments because these are the situations where
automatic processes will have dominating influences.
- Kenny suggested that we might try collecting some pilot data
using the PY101 pool. One useful topic might be dating skills. This
topic is relevant to most of the population, and the performance
probably has a strong automatic component to it.
- Forrest mentioned some work by Segal that might be relevant.
One article is available on this website.
- The following suggestions were made as to profitable directions
to pursue at this point in time.
- Collect pilot data with theraputic implications.
- Consider writing an R03 or an R34
- Talk to the clinic about regularly collecting implicit
measures on clients
- Have a group meeting with Steve Holland (a big name in CBT)
when he comes to Alabama in March.
- Contact researchers already known to be investigating similar
topics: Michael Vasey, Bethany Teachman
- February 8, 2006
- Meeting attendees (specifically for cog folks): Jamie D,
Mark, Joe, Jamie O, Chris
- We started with a discussion of the Gemar, et al. (2001)
article. While we thought it was very promising in that it indicated
that clinical psychologists would be receptive to the use of implicit
measures, we noticed a number of methodological flaws.
- The IAT results were broken down by the type of response
("me" vs. "not me"), while the standard IAT just looks at consistent
vs. inconsistent. This implies that the overall results may not have
been significant.
- The reported IAT effect don't use Greenwald's new measure (d)
which corrects for the overall response time. Looking at the pattern of
results we believe the strength of the effects would be weaker if they
used the more appropriate measure.
- The authors emphasize that significant results were found in
the "me" condition but not in the "not me" condition, but the pattern
found in both are fairly similar.
- When interpreting the results the authors don't seem to
appreciate that implicit and explicit cognition may result from
different processes.
- We next talked about choosing measures that might be useful to
give to people when they enter and exit the psychological clinic. We
wanted to focus on well-established measures that would most directly
be relevant to people entering the clinic. Broadly, we plan to
investigate construct accessibility, implicit self-esteem, and
positivity bias. We plan to use the following measures.
- Attitude accessibility (for things most relevant to
psychological problems, such as relationship words, job
words, work words, and self-related
words.
- Implicit
self esteem
- Stroop
task for negativity bias
- Baldwin's
social search task to detect attention to affect
- Self-esteem
IAT
- Our plan is to first test a bunch of intro psych students on
these measures across two time points and see which are stable over
time. Measures that don't represent stable psychological constructs
will not be likely to give us any useful information.
- Each of the attendees agreed to get more details on one of
these measures. We plan to meet again once everyone has had a chance to
investigate their topic. The assignments are:
- Jamie D- attitude accessibility
- Jamie O - Implicit self-esteem
- Joe - Stroop
- Chris - Baldwin's social search task
- Mark - Self-esteem IAT
- March 7, 2006
- Meeting attendees (again, specifically for cog folks):
Jamie D, Mark, Chris, Jamie O
- Jamie D suggested (based on conversations with Heather Claypool
and Anne-Marie Leistico) that we additionally consider the mere
exposure task to provide a measure of an individual's resistance to
change.
- Obtained specific references for each of the implicit measures
that we are considering. These articles are posted on the website in
the readings section above.
- Jamie O provided an article by Bosson that reports the
test-retest reliability of several measures of implicit self-esteem.
The only two that seem to work well are the Name letter effect and the
IAT.
- We should collect a measure of explicit self-esteem, since some
prior research has indicated that the interaction between implicit and
explicit self-esteem can be particularly important. Two possible
measures of explicit self esteem would be the Rosenberg self-esteem
scale (which Joe used in his masters) and the Beck Depression Inventory.
- Baldwin's social search task originally looked to see how
performing this task influenced later performance on a Stroop task.
However, we will probably just want to just look at performance on this
task separately as a measure of positivity/negativity bias. We may also
want to vary the method so that on some trials participants are
searching for positive faces while on others they are searching for
negative faces. At the beginning of each trial we can have participants
click a work in the center of the screen that tells them which type of
face they will be searching for. This not only guarantees that they
know what they are supposed to do, but it also brings the mouse to the
center of the screen at the beginning of each trial. We will probably
want to use real faces - Mark mentioned that he has a database of faces
with different characteristics that we might use. Jamie D also has a
large number of photos from an online photo archive.
- March 10, 2006
- Meeting attendees: Steve Hollon (guest speaker with a strong
familiarity with cognitive therapy), Jamie D, Mark, Forrest, Kenny,
Martin
- Steve mentioned the fact that the types of things we examine
with our implicit measures (schemas, construct accessibility,
underlying associations, etc.) are actually talked about a great deal
in cognitive therapy, even though therapists currently don't have any
good ways of measuring them.
- Steve mentioned that after people have a depressive episode,
they remain more vulnerable to having future episodes even after their
emotional standing has returned to normal. A new event can bring them
right back to where they were when they first needed therapy. This
implies that there is something that has specifically not been changed
in the person, even though they show improvement on measures of
depression.
- We might want to consider using pharmacotherapy as a control
group. Pharmacothereapy really should have no ability to change any
underlying cognitive structures even though it is often shown to be
just as effective at reducing symptoms as psychological therapy.
Evidence for this is found in that CBT seems the reduce the probability
of future setbacks relative to pharmacotherapy, even though both appear
to perform equally well in reducing symptoms of depression.
- Early phases of therapy appear to require explicit effort, but
techniques become more automatized over time.
- Different areas of the brain appear to be affected by
pharmacotherapy and CBT. Research on this is being performed by Segal
(neuroscientist) and Thase (clinician) at Pittsburgh, and Mayberg at
Emory.
- Steve was involved in the development of an "Automatic Thoughts
Questionnaire," but he feels that this doesn't truly capture
automaticity. Instead, it captures those things that actually *are*
consciously available to the respondent. Gets more at spontaneous
thoughts than implicit associations/accessibility.
- To uncover what implicit associations an individual has, a
therapist will first ask a client to imagine a situation and then ask
"What would it mean if..." (some negative event). This then would lead
to a new event, about which a the therapist could ask another "What
would it mean if..." question. Tracing things down you can usually
determine what underlying negative automatic responses the client has.
You get different "trajectories" from clients who have good versus poor
coping skills.
- Differences at the implicit level likely have their greatest
influence once a client is actually no longer having an active episode,
determining their vulnerability to new episodes.
- We likely don't need to worry about randomly assigning people
to pharmacothereapy/psychotherapy groups when we are examining
assessment. We could therefore simply try to connect with a
pharmacotherapist and just ask them to have their clients fill out our
implicit measures before and after treatment, and then do the same with
similar patients going to a psychologist. We will, however, need to
worry about random assignment if we want to examine a treatment.
However, this itself brings up possible ethical/practical issues of
failing to provide optimal care for clients. Existing institutions will
be very resistent to methodological innovations. Sometimes
adminstrators will violate the random assignment to try to help people
as much as possible. It might be easiest to have a design comparing
pharmacological treatment + ICT treatment to pharmacology treatment
alone.
Links