Implications of Developments in Cognitive Neuroscience for ...

Text-only Preview

Editor: Drew Westen, PhD
Implications of Developments in Cognitive
Neuroscience for Psychoanalytic Psychotherapy
Drew Westen, PhD
During the 1960s and 1970s, psychology underwent a cog-
conscious processes are understood to be superimposed upon
nitive revolution in which information-processing mod-
a more pervasive parallel (many-at-a-time) architecture.
els based on the metaphor of mind-as-computer supplanted
Thus, when we perceive tears in a patient’s eyes as he talks
the behavioral models that had dominated experimental
about his divorce, we are processing visual information from
psychology for much of the previous century. In similar fash-
his face simultaneously with the sound and meaning of his
ion, we are currently witnessing what might be called a sec-
words. Together, what we see and what we hear lead to an in-
ond cognitive revolution in experimental psychology and
stantaneous conscious perception of tears. Had the barely
the neurosciences, characterized by a number of interrelated
perceptible increase in the amount of liquid surrounding his
features. In this article I describe a few of the central features
eyes not been accompanied by his words and our under-
of this second cognitive revolution and then briefly outline
standing of their meaning to him, we might not have noticed
some of the implications and challenges for psychoanalytic
it at all or might have interpreted it differently.
A second feature of the current cognitive revolution is a
conceptual shift from memory stores to memory systems. In
this view, knowledge resides not in “places” or “engrams” in
the brain, but in distributed networks of neurons whose co-
activation constitutes a representation. To put it another way,
The first feature of this emerging paradigm shift is a change
when we perceive, say, a tear, a network of neurons is acti-
in emphasis from serial to parallel processing models. In se-
vated simultaneously. When we remember that tearful look
rial processing models, units of information come into con-
(and perhaps the narrative that accompanied it), enough of
sciousness one at a time (serially), then either fade away or
the original network is reactivated to represent—literally, to
become stored in long-term memory. From long-term mem-
re-present—the experience in our minds. This network is dis-
ory, they can be retrieved into short-term memory, which
tributed (that is, dispersed) over a large number of neural
holds roughly seven pieces of information (such as seven dig-
units, whose simultaneous activation constitutes a memory.
its) for about 30 seconds.
Thus, in this view, a memory is not a “thing” that is stored
In parallel processing models, these serial, one-at-a-time
somewhere in a mental warehouse and can be pulled out
and brought to the fore. Rather, it is a potential for reactiva-
of a set of neurons that together constitute a particular
*For a more extensive discussion of these issues, see Westen and Gab-
bard.1,2 For examples of this second cognitive revolution in the
A third feature is the increasing recognition of the impor-
cognitive neuroscience literature, see Rumelhart and McClelland,3
tance of distinguishing between explicit memory (conscious
Schacter,4 Smith,5 and Squire.6 For recent examples of integrative
recall or recognition, such as remembering a prior experience
efforts linking psychoanalysis and cognitive science, see Bucci,7
or the name of a medication) and implicit memory (memory
Horowitz,8 Olds,9 Shevrin et al., 10 and Westen.11
outside of conscious awareness). Of particular relevance for
psychotherapy is the distinction between two kinds of mem-
Reprint requests: Drew Westen, PhD, Departments of Psychiatry and
ory that are often implicit: associative memory and proce-
Psychology, Emory University, 532 N. Kilgo Circle, Atlanta, GA
dural memory. Associative memory involves the unconscious
activation of networks of association—thoughts, feelings,
Harvard Rev Psychiatry 2002;10:369–373.
wishes, fears, and perceptions that are connected, so that ac-
tivation of one node in the network leads to activation of the
© 2002 President and Fellows of Harvard College
others. Implicit associative memory can be seen in priming

Harvard Rev Psychiatry
November/December 2002
experiments. For example, presentation of the word “taxi”
ous activation of information-processing units that together
renders an individual more likely to use the less common
could constitute a unit of meaning and for the way in which
spelling of the homophone fare/fair, because “taxi” primes the
the “firing” of one unit could alter the likelihood of activation
associated word “fare.”
of other associated units. The shift in metaphor is of particu-
Procedural memory refers to skills that are expressed
lar significance because it brought into cognitive models three
without conscious awareness and that may even be disrupted
phenomena that were shadowy figures in the cognitive sci-
if the person tries to perform them using conscious control.
ence of eras past: emotion, motivation, and consciousness.
Skilled musicians, for example, can produce notes far faster
Computers lack all three, but brains do not. For practicing
than they can consciously read them, just as skilled psy-
clinicians, the shift could not be more crucial, because infor-
chotherapists can respond to complex, multidetermined com-
mation-processing models that cannot address emotional
munications from their patients long before they can con-
phenomena, wishes and fears, and information processed at
sciously process the multiple levels of meaning. Indeed, one of
different levels of consciousness can be of only limited use
the challenges of clinical supervision is learning to verbalize
to them.
knowledge that is procedural—to make the implicit explicit.
Juxtaposed with procedural knowledge is declarative knowl-
edge—knowledge of facts (events, abstractions, details) that
can be “declared.”
Although researchers often use the words “implicit” and
The relationship between psychoanalysis and developments
“explicit” as synonyms for procedural and declarative knowl-
in psychology, psychiatry, and the empirical sciences more
edge, the implicit/explicit distinction refers to the way in
generally has always been ambivalent. Freud usually con-
which information is remembered (unconsciously or con-
sidered psychoanalysis a science (to this day, paper presen-
sciously), whereas the procedural/declarative distinction in-
tations at the meetings of the American Psychoanalytic As-
volves the manner in which knowledge is represented (as a
sociation are listed on the program under the heading of
skill or as a fact encoded in a linguistic or sensory mode). De-
“scientific proceedings”), but he wrestled throughout his ca-
clarative knowledge can be implicit (such as words that are
reer with the difficulties inherent in constructing a science of
primed and hence activated but not currently within one’s
awareness, or incidents that are too painful to remember ex-
In many respects, recent developments in the cognitive
plicitly), and procedural knowledge can be explicit (such as
neurosciences have breathed new life into psychoanalytic
coping strategies, which are skills for regulating emotions).
forms of psychotherapy, particularly long-term therapies, pro-
Indeed, one of the ways in which cognitive neuroscience
viding them an empirical grounding in basic science even if
might benefit from greater contact with psychoanalysis is in
they lack solid grounding in applied research on treatment
addressing phenomena such as implicit declarative thoughts
or memories (e.g., implicit knowledge about negative aspects
A testimony to the increasing convergence of psychoana-
of a loved one that are kept from conscious awareness because
lytic ideas with concepts from the cognitive neurosciences is
they would elicit unpleasant feelings).
the number of articles in the psychoanalytic literature that
A fourth feature of this second cognitive revolution is a
reference concepts such as implicit and explicit memory and
growing awareness of the role of emotion and motivation in
procedural knowledge. In particular, the second cognitive rev-
thought and memory and, with it, an increased focus on real-
olution promises to provide an empirical grounding for two
life, meaningful memory (rather than memory stripped of
central psychoanalytic concepts: unconscious associative net-
meaning in the laboratory, such as pairing of nonsense syl-
works and unconscious procedures. A century after Freud’s
lables). Researchers are increasingly studying phenomena
pioneering clinical/theoretical efforts, a virtual explosion of
ranging from traumatic memories and the impact of trauma
experimental research into unconscious (implicit) processes
on the brain to more motivationally significant processes,
has borne out his most fundamental assumption—namely,
such as how we remember to remember to do something later
that unconscious associative networks regulate much of what
in the day.
we think, feel, wish for, fear, and do.11,12 In suggesting that
Finally, and perhaps central to all of these features, is a
much of our mental life is organized around unconscious as-
shift in metaphor, from mind-as-computer to mind-as-brain.
sociational networks, Freud drew on the same British asso-
Whereas cognitive scientists of the 1960s drew their inspira-
ciationist tradition employed by contemporary cognitive sci-
tion from the high-speed computers of their day, those of the
entists. If it is nothing else, psychoanalysis as a treatment is
1980s and 1990s were equally inspired by developments in
a process aimed at trying to map a patient’s idiosyncratic as-
the neurosciences. Thus, the simultaneous activation of a net-
sociative networks, particularly those that appear relevant
work of neurons and the transmission of neural information
to symptomatic sources of distress, in the process giving the
at the synapse became useful metaphors for the simultane-
patient more freedom to make conscious, explicit choices.

Harvard Rev Psychiatry
Volume 10, Number 6
Classic psychoanalytic models of conflict and compromise
chotherapy. Free association is one of the pillars—indeed, the
further emphasize the extent to which we carry out complex
fundamental rule—of psychoanalysis proper as a technique,
procedures to regulate emotions (defenses) and balance com-
predicated on the assumption that free association is rela-
peting motivational and emotional pulls (compromise for-
tively free except when encumbered by defense. A cognitive-
mations). Object relations and contemporary relational
neuroscientific account challenges the psychoanalytic view
approaches to psychoanalysis emphasize implicit represen-
of pristine, unfettered associations originating in the patient’s
tations of self and relationships as well as the implicit rela-
childhood or even in the relational matrix of the analytic
tional procedures that guide much of the way we think, feel,
dyad. What is on a person’s (unconscious) mind at any time
and interact in intimate relationships. Although neuro-
is a joint function of what is chronically there (much of which
scientists have not yet integrated these kinds of complex af-
is likely, in fact, to reflect concerns originally forged in child-
fective, motivational, and interpersonal/intrapsychic phe-
hood) and what has been there recently, which may or may
nomena into their models, the features of the second cognitive
not be related to the concerns that brought the individual
revolution described above make such an integration much
to treatment. Recency effects include not only the events of
more possible. Indeed, one of the major ways in which psy-
the analytic hour but also extraanalytic events that could be
choanalysis can contribute to cognitive neuroscience, rather
important to explore. The particular associations that emerge
than just learn from it, is to bring to bear clinical data and
in any analytic hour may or may not be useful to examine;
conceptualizations that would probably not emerge from the
with respect to core dynamics, they may represent signal or
laboratory, where the focus is on nonidiosyncratic networks
noise. Over time one would expect important material to be
that lead subjects to respond similarly to an experimental
reflected repeatedly in the patient’s associations, particularly
if the analyst makes few interpretations that shape these as-
Although psychoanalysts have reason to rejoice in the find-
sociations, since chronically activated networks produce de-
ings of contemporary research in cognitive neuroscience, this
rivatives in both the conscious thought and the associations
research also poses some substantial challenges.1,2,12 First
of the patient. However, this is likely to be a very inefficient
and foremost, data from the cognitive neurosciences suggest
process (and indeed is what drove many of the early cognitive
that many unconscious processes are unconscious not by
therapists, such as Beck and Ellis, to leave the analytic fold),
virtue of defense but because of the construction of the hu-
and the advantages of keeping the patient’s associations rel-
man brain. People are often unaware of their implicit net-
atively free of the analyst’s theories and interests—even if
works of association because these networks are, in Freud’s
doing so were possible—come at the cost of other potentially
terms, descriptively but not dynamically unconscious. We
important interventions.
have no more access to our implicit associational networks
To complicate matters, patients can (consciously and un-
than we do to the perceptual processes by which we locate an
consciously) veer away from entire networks of association
object in space. The same is true of many implicit relational
that make them feel anxious, guilty, or ashamed, potentially
procedures that are the center of a substantial amount of clin-
keeping central issues out of the therapeutic spotlight for
ical work—relational procedures forged in a matrix of tem-
years, because doing so provides short-term relief from un-
perament, experience with early attachment figures, and
pleasant feelings. Further, a patient who avoids feared activ-
evolving relationships over time.
ities (for example, an individual who avoids meaningful re-
None of this is to deny the importance of defensive
lationships due to fear of intimacy) is unlikely to have the
transformations of unconscious representations—motivated
relevant networks at a high enough state of activation to ex-
decision-making that leads people to see themselves, oth-
plore without actual exposure to the feared situation. Thus,
ers, or situations in a light that minimizes negative affect
the analyst may need to depart from neutrality vis-à-vis the
and maximizes positive affect, or motivated failure to become
patient’s conflicts and encourage activities that both expose
conscious of activated representations that are threatening.
the patient to what he or she has been avoiding (which has
The point is simply that making the unconscious conscious
the potential to foster the development of new networks, a
is more complex than it may once have appeared, because it
central aim of behavioral treatments) and that bring to the
entails not only helping people to see what they have been
fore the most important associative material. Freud was the
afraid to see but also helping them to see patterns of func-
first analyst to practice this way, noting at one point that
tioning—ways of dealing with emotions, of interacting with
people cannot free themselves of their fears unless they con-
other people, of viewing themselves—that they could not
front them. The dangers in “taking sides” with respect to a
have seen even if they had wanted to.
patient’s conflicts lie in making an internal conflict an exter-
What are the implications for psychoanalytic psycho-
nal one between patient and therapist, and in the potential
therapy? Perhaps the major implication relates to free as-
for therapists or analysts to act on the basis of their own coun-
sociation, the use of which constitutes a major technical
tertransferential pulls rather than in the patient’s best in-
distinction between psychoanalysis and psychoanalytic psy-
terest. This paradox has no solution, of course, but good treat-

Harvard Rev Psychiatry
November/December 2002
ment probably requires a balance of exploration and exhor-
described by numerous psychoanalytic theorists) or more
tation in the service of further exploration and behavioral
idiosyncratic (the “old” meaning of countertransference,
as a problematic intrusion of the therapist’s issues into the
Although it is not free of limitations, free association is
clearly one of the most effective ways of learning about a pa-
Similarly, the notion of representations as potentials for
tient’s associative networks. Time constraints, however, tend
reactivation offers supportive data at the level of basic sci-
to limit its use in the kinds of weekly or twice-weekly psy-
ence to the long-held psychoanalytic position that patients
chodynamic psychotherapies widely practiced in the commu-
are likely to express important conflicts, defenses, motives,
nity. One useful way to employ associative techniques in psy-
and relational patterns in their interactions with the thera-
chodynamic psychotherapy is to use a form of directed free
pist or analyst. From a cognitive point of view, the therapeu-
association, departing at various points from the more con-
tic situation, to the extent that it matches prototypes from
versational style of treatment and asking for the patient’s as-
the past, is likely to activate such responses.16,17 Thus, it
sociations to important material or significant memories as
should not come as a surprise if important relational patterns
a way of exploring the underlying associative networks. For
emerge in a situation in which one person deeply self-
example, one patient who repeatedly “dumped” women as
discloses and becomes attached in an intimate but asymmet-
soon as their relationship became sexual had no idea why he
rical relationship with another who is trying to be helpful,
was doing this, although he was always ready with an expla-
nurturant, and attentive to that person’s needs. Simply by
nation. On one occasion, after the pattern became clear, I
virtue of the way the therapeutic frame is set, the therapist is
asked him to imagine as vividly as possible the events of the
likely to be experienced as an authority figure, an attachment
prior evening, particularly the events that occurred in bed
figure, an object of love, an object of desire, and so forth.
before his latest paramour took her final leave. I asked him to
At the same time, research in cognitive neuroscience chal-
walk me through the episode moment by moment, eliciting
lenges a number of views of transference once taken as ax-
associations along the way, suggesting that he report any-
iomatic but increasingly challenged in the analytic literature,
thing he thought, felt, saw, tasted, smelled, or felt during the
such as the view that everything important about the pa-
encounter and any memories or images that emerged in re-
tient’s psychopathology will ultimately come out in the ther-
calling it. My aim was to activate a multimodal network and
apeutic relationship or that characteristics of the patient’s
to prime as many of its elements as possible. Doing so led to a
internal relational paradigms are likely to express them-
series of associations about his fears of his own aggression
selves regardless of features of the analyst such as age, gen-
and his tremendous guilt at these impulses and fantasies—
der, or style. Although some relational paradigms are likely to
and to several weeks in which we did virtually nothing but
be so tenacious that they will ultimately turn up in virtually
associative work of this sort, after which the conflict resolved
every significant relationship, including a therapeutic one,
itself and never returned.
thinking about the conditions for activation of representa-
Data from the cognitive neurosciences may also help to
tions should lead to clearer thinking about the circumstances
shed new light on classic psychoanalytic constructs such as
in which important material is likely to “emerge in the trans-
object representations, transference, and countertransfer-
ference” or will require examination of extra-transferential
ence. Viewing object representations (representations of the
self and others) as potentials for reactivation of previously
active neural networks that include both declarative and pro-
cedural elements may lead psychoanalytic theorists to be
more precise in their models of how representations form and
Data from the cognitive neurosciences are not likely to revo-
guide relational functioning. The notion that a patient can,
lutionize the practice of psychotherapy, but they are certainly
for example, “project into the therapist” a particular repre-
likely to enrich it. At minimum, these data corroborate the
sentation or way of responding requires much more careful
importance of attending to unconscious (implicit) networks
explication than is common in psychoanalytic literature,
that regulate many of the ways we think, feel, desire, fear,
where too often clinically rich description alloyed with
and fall ill. A fundamental question at this point is how to im-
metaphor passes for explanation (see Westen, in press15).
prove our technologies for changing those networks.
Such an explication would require theorists to be more spe-
cific about how particular representations or procedures ac-
tivated in the patient become expressed implicitly and how
they trigger particular patterns of activation in the thera-
pist that are either culturally patterned (and hence lead
1. Westen D, Gabbard GO. Developments in cognitive neuroscience,
to ways of responding that virtually anyone would have,
1: Conflict, compromise, and connectionism. J Am Psychoanal
producing the kinds of repetitive interpersonal patterns
Assoc 2002;50:54–98.

Harvard Rev Psychiatry
Volume 10, Number 6
2. Westen D, Gabbard GO. Developments in cognitive neuroscience,
10. Shevrin H, Bond JA, Brakel LA, Hertel RK, Williams WJ. Con-
2: Implications for theories of transference. J Am Psychoanal As-
scious and unconscious processes: psychodynamic, cognitive, and
soc 2002;50:99–134.
neurophysiological convergences. New York: Guilford, 1996.
3. Rumelhart DE, McClelland JL. Parallel distributed processing:
11. Westen D. The scientific legacy of Sigmund Freud: toward a psy-
explorations in the microstructure of cognition, vol 1: Founda-
chodynamically informed psychological science. Psychol Bull
tions. Cambridge, Massachusetts: MIT Press, 1986.
4. Schacter DL. Understanding implicit memory: a cognitive neu-
12. Westen D. The scientific status of unconscious processes: is Freud
roscience approach. Am Psychol 1992;47:559–69.
really dead? J Am Psychoanal Assoc 1999;49(suppl):1–30.
5. Smith ER. Mental representation and memory. In: Gilbert DT,
13. Wachtel PL. Psychoanalysis and behavior therapy: toward an in-
Fiske ST, Lindzey G, eds. The handbook of social psychology, vol
tegration. New York: Basic, 1977.
1. 4th ed. New York: McGraw-Hill, 1998:391–445.
14. Wachtel PL. Action and insight. New York: Guilford, 1987.
6. Squire LR. Declarative and nondeclarative memory: multiple
15. Westen D. The language of psychoanalytic discourse. Psychoanal
brain systems supporting learning and memory. J Cogn Neurosci
Dialogues [In press].
16. Andersen SM, Baum A. Transference in interpersonal relations:
7. Bucci W. Psychoanalysis and cognitive science: a multiple code
inferences and affect based on significant-other representations.
theory. New York: Guilford, 1997.
J Pers 1994;62:459–97.
8. Horowitz MJ. Cognitive psychodynamics: from conflict to char-
17. Luborsky L, Crits-Christoph P. Understanding transference: the
acter. New York: Wiley, 1998.
Core Conflictual Relationship Theme method. 2nd ed. Washing-
9. Olds DD. Connectionism and psychoanalysis. J Am Psychoanal
ton, DC: American Psychological Association, 1998.
Assoc 1994;42:581–611.