Creative Education
2013. Vol.4, No.6A, 1-8
Published Online June 2013 in SciRes (
Copyright © 2013 SciRes. 1
Mechanism of Discerning Similarities in Psychiatric Diagnoses
Haim Eshach
Ben Gurion University of t he Ne ge v , Beer Sheva, Israel
Received April 23rd, 20 1 3; revised May 23rd, 2013; accepted May 3 1st, 2013
Copyright © 2013 Haim Eshach. This is an open access article distributed under the Creative Commons Attribu-
tion License, which permits unrestr ic ted u se, distribution, and r eproduction in any medium, provided the original
work is properly cited.
Psychiatrists use similarities to make diagnoses in situations where rules are insufficient. The purpose of
this study was to examine the criteria psychiatrists use to determine similarity between cases of personal-
ity disorder. Psychiatrists were provided with cases and were interviewed according to Kelly’s method.
Cognitive science theories were used to explain the criteria identified. Results indicated that similarity de-
cisions include idiosyncratic criteria. These results add to the current understanding of medical prob-
lem-solving processes and have implications for medical education.
Keywords: Similarity; Psychiatric Diagnoses; Medical Education
Clinical reasoning, a highly complex phenomenon (Higgs &
Jones, 1995), is defined as the thinking processes occurring
while dealing with a clinical case (Boshuizen & Schmidt, 1995).
The problems that patients present can be confusing and con-
tradictory, characterized by imperfect, inconsistent, or even
inaccurate information (Kassirer & Kopelman, 1991; Eshach &
Bitterman, 2002). In addition, medical problems are often
poorly defined (Barrows & Feltovich, 1987). Not only is much
irrelevant information present, but relevant information about
the case is often missing and does not become apparent until
after problem solving has begun (Voss & Post, 1988). In such
cases, it is difficult to specify the state from which one can start,
to identify the medical interventions that might be applicable,
or even to recognize when the goal has been achieved. In other
words, many medical problems are ill-defined. Moreover, in
many situations, the deterministic mechanisms that account for
the medical problems are not completely understood (Williams,
1992). In other words, many medical domains are “ill-domain
theories”. Despite the uncertainty and variation that character-
ize physicians’ work they have to diagnose and treat such ill-
nesses. In order to do this, physicians use different types of
reasoning e.g. rule-based reasoning and similarity-based rea-
soning. In the following paragraphs I first explain why person-
ality disorder realm was chosen for the purpose of this study.
Secondly I explain why rule-based reasoning cannot alone ex-
plain the variety of medical decision making processes. Third, I
argue that similarity decisions may be effective to solve
ill-defined problems in ill domains. I then suggest to study the
factors according which similarities are made.
Personality Disorders as the Focus of the Study
The etiology and pathogenesis of most personality disorders
are not known; psychiatrists therefore tend to resort to phe-
nomenological, descriptive criteria in the diagnosis of these
conditions. Many of the definitions provided in the Diagnostic
and Statistical Manual of Mental Disorders (DSM-IV) (Ameri-
can Psychiatric Association, 1994) are set up in a format
whereby an individual must have a subset of several features in
order to be diagnosed. Thus, for the nine DSM-IV diagnostic
features of borderline personality, for instance, at least five of
which are required for diagnosis, there are 256 different criteria
combinations that can result in this diagnosis (Morey, 1998). In
addition, the study of personality disorder leads into areas in
which the distinction between health and sickness is inherently
ambiguous and is potentially biased by societal norms and per-
sonal values. To a large extent, whether persons are considered
to have a personality disorder depends on whether their person-
ality traits are noxious to the society in which they live and, in
particular, to the persons with whom they most frequently in-
teract (Kaplan & Sadock, 1996). This means that formal diag-
nostic systems such as the DSM-IV, which define rules to cate-
gorize discrete pathological entities, are not sufficient and can-
not always lead to the diagnosis of the personality disorder.
Despite the uncertainty and variation that characterize personal-
ity disorders, psychiatrists have to diagnose and treat them. It is
my assumption that perceiving similarities, an ability that is one
of the most fundamental aspects of human cognition (Vos-
niadou & Ortony, 1989), is used by psychiatrists in the diag-
nostic and treating of personality disorders.
Domain Rule -Based Rea soning
In order to better understand this phenomenon one should
first distinguish between domain rule-based reasoning (RBR)
and the use of a rule or rules in problem solving. The former is
identified as the process of drawing conclusions by chaining
together generalized rules, starting from scratch (Leake, 1996).
The latter, on the other hand does not require starting from
scratch nor does it require chaining together rules. For the pur-
pose of this study I will focus on RBR processes. RBR models
are rooted in the philosophical belief that humans are rational
beings and that the laws of logic are the laws of thoughts (Ey-
senck & Keane, 1995). According to Kolodner (1993), although
some rules are very specific, the goal is to formulate rules that
are generally applicable. An important advantage of rules is the
economy of storage they allow. But general knowledge has the
problem of applicability (Mostow, 1983), i.e., the bringing of
some general piece of knowledge to a particular situation.
When rules are expressed too abstractly, the terms tend to be
unintelligible to the novice and to mean a variety of specific
things to the expert. Also, in ill-defined domains, the rules do
not encompass all the situations they are asked or assumed to
cover, admit tacit exceptions, or can be contradicted and an-
nulled by other rules (Rissland & Skalak, 1991).
Limited mental capacity is another reason why RBR alone
cannot explain the human problem-solving processes. RBR,
which requires that the problem solver take into account all the
domain rules, is not capable of doing that in real-life situations
(BarOn, 1993), as the number of rules required for solving a
problem may be unmanageably large (Leake, 1996). Similarity
based reasoning offers an attractive way out of these difficulties.
Rather than having to patch up abstract rules with endless
sub-rules, according to similarity-based models of reasoning,
problem solving depend, especially while working in ill-do-
mains theories and ill problems, on the ability to identify the
most relevant bodies of knowledge that already exist in mem-
ory so that this knowledge can be used as a starting point for
encounter the novel situation. I do not focus on a particular
model such as the prototype models, the exampler models,
connectionist models and the probabilistic models, but rather
relate to the similarities underlying all these kinds of models.
Rumelhart (1989) claims that most everyday reasoning
probably involves assimilating the novel problem to other
situations in which the solutions are known—that is, reasoning
by similarity. He further suggests that the reason for the impor-
tant role of reasoning by similarity is the essential human abil-
ity for pattern matching. “We seem to be able to ‘settle’ quickly
on interpretation of an input pattern. This is an ability that is
central to perceiving, remembering, comprehending, and rea-
soning by similarity. Our ability to pattern—much is probably
not something that sets humans apart from other animals but is
probably the essential component to most cognitive behavior”
(Rumelhart, 1989: p. 300). Support for similarity processes can
be found in cognitive neurosciences research. This indicates
that different neural circuits are involved when people catego-
rize items on the basis of a rule, as compared with when they
categorize the same items on the basis of similarity (Kolodny,
The term “similarity”, which typically refers to the outcome
of comparison among entities, can be regarded as a form of
judgment—what people say when asked to compare different
entities (Sloman & Rips, 1998). Goodman (1972) argues that
claiming that two things are similar is uninformative until we
specify in what respect they are similar. According to Gold-
stone (1994) people do not usually compare objects only in a
single respect such as “size” but along multiple dimensions
such as size, color, shape, etc. Kelly (1955a) claims that two
things are “similar”, “alike”, or “identical” obviously means
that they are alike in some particular way or ways, but, of
course, never in every way. Their alikeness makes no sense
unless it also serves to distinguish them from certain other
things. Thus, according to Kelly, likeness always implies a
difference. At the same time, the way in which two things are
different must, if it is to make any sense at all, be the way in
which at least one of them is like a third thing.
This study focused on understanding the process of how pro-
fessionals find similarities between cases. The following ques-
tions are therefore addressed:
1) What are the differences between the various diagnoses
given by psychiatrists to given cases?
2) What are the criteria that psychiatrists use to decide on
similarities between cases?
Nine psychiatrists at different levels of medical experience
were chosen for this study.
Participant 1: 20 years’ experience; Participant 2: 6 years’
experience; Participant 3: 7 years’ experience; Participant 4:
10 years’ experience; Participant 5: 5 years’ experience; Par-
ticipant 6: 2 years’ experience; Participant 7: 4 years’ experi-
ence; Participant 8: 5 years’ experience; Participant 9: 5 years’
1) Nine patients diagnosed as suffering from personality dis-
orders according to DSM-IV criteria who agreed to participate
in the study were identified. The following list describes the
diagnosis given by the department in which the patients were
a) Narcissistic personality disorder with dependent traits;
b) Borderline personality disorder with narcissistic traits;
c) Narcissistic personality disorder;
d) Borderline personality disorder;
e) Narcissistic personality disorder;
f) Dependent personality disorder;
g) Narcissistic personality with obsessive traits;
h) Obsessive-compulsive disorder;
i) Borderline personality with narcissistic traits.
A psychiatrist who was very familiar with all the cases con-
ducted an introductory interview with each one of the patients.
This enabled him to extract the maximum and most pertinent
information from each case. The interviews lasted between 20
and 25 minutes. The interviews were videotaped and tran-
scribed. Transcripts of each interview were used as cases in-
troduced to the participants.
2) The transcripts of the nine cases were presented to each of
the participants, all of whom were unfamiliar with the nine
patients. The participants were each instructed to do the fol-
a) Diagnose each case. The participants were asked to think
aloud while diagnosing the cases. These sessions were tape-
recorded, and verbatim transcripts were produced.
b) Group cases according to important similarities of their
own choice.
1This idea is the basis for Kelly’s (1955) repertory test, from which we have
drawn the interviews described in the method for the current research.
Copyright © 2013 SciRes.
c) Consider predefined sets of three cases and describe how
two of the three cases are alike and how they are different from
the third; in other words, to decide on a criterion for the simi-
larity between two cases among the three presented in the set.
d) Subsequently, to indicate which of the remaining six cases
matched the chosen criteria.
For c) and d), Table 1 was used as a template for the inter-
The author interviewed each participant, each interview last-
ing between 5 and 6 hours. Since the interviews were so long,
they were divided into two parts of between 2.5 and 3 hours
each, conducted on two successive days. In this way the par-
ticipants remembered the cases, and it took very little time for
them to go over the cases again. After piloting the interview
method, it became clear that 14 sets were optimum for inter-
view acceptability by the participants.
This method required the identification of the two most
similar cases of a set of three and enabled us to identify the
variety of criteria used to decide on similarities between cases.
For instance, when participant 4 was introduced to set 4 (cases
2, 5, and 8), she identified cases 2 and 8 as similar. She ex-
plained, “In both cases they try to solve an internal conflict.
The crisis starts early in life. Case 5 is not trying to solve an
internal conflict”. When the participant was presented with set
5, she again identified cases 2 and 8 as similar. She explained
that both cases had “organized life, while case 3 had chaotic
life”. For the same pair of cases she used different criteria.
The interview audiotapes were transcribed verbatim, and in-
ductive analysis, in which patterns, themes, and categories of
Table 1.
Interview template.
number Set of three
casesa Similar
casesb Expl an atio n c Other similar
1 2 3 4
2 2 3 5
3 3 5 7
4 2 5 8
5 2 3 8
6 1 6 8
7 1 3 5
8 4 5 9
9 2 4 5
10 2 6 8
11 5 7 9
12 2 4 9
13 4 5 8
14 3 7 9
Note: aPredefined sets of three cases each. bThose two cases that the participant
identified as similar. cReasons that the participant gave in answer to “in what respect
are the cases similar?” dThose cases of the six remaining that the participant
identified as similar to the earlier two cases already chosen for their similarity.
analysis derived from the data (Patton, 1990), was employed.
The analysis process contained the following:
All the transcripts were repeatedly read to formulate tentative
a) In subsequent readings, an attempt to find confirmation or
disconfirmation to the tentative understanding of the phenom-
ena on the tape was made.
b) Through this process of constant comparison (Strauss,
1987), initial categories for the differences between the diag-
noses were established.
c) In the same manner, criteria used to decide similarities
were identified, and initial categories for the criteria were rec-
d) The data were repeatedly reread, and the initial categories
were revised as a result of several rounds of discussion verifi-
cation methodology (Strauss, 1987).
e) This sharing and critiquing also assisted in the process of
progressive subjectivity (Guba & Lincoln, 1989).
Results and Discussion
Results Regarding the Diagnoses
1) Differences between the diagnoses given by the nine par-
ticipants were found. Table 2 summarizes the diagnoses that
were given to each case by the participants.
2) Four classes of difference were identified. See Table 3.
3) The data indicated that the participants focused on defer-
ent details in each case. The following citations of three par-
ticipants (2, 6, and 9) while diagnosing the same case (6) dem-
onstrate this phenomenon:
Participant 2: “He [the patient] describes his life through his
functional ability and less through interrelationships with others.
Despite not being uncomfortable around people, he was never
close to them. There is a personality disorder on the basis of
creating relationships with other people … In this case schizoid
personality disorder seems to me as the one he suffers from …”
Participant 9: “The more reasonable possibility seems to me
that he wants the insurance money. He does not describe psy-
chiatric details, but things that anyone can describe. He contin-
ues to describe details from the accident and what happened
due to the accident. He doesn’t cooperate on discussing details
after the accident. He does not remember how he was as a high
school student, which indicates memory disorder, which seems
unreasonable in this case.”
Participant 6: “His work provided him with high self-esteem.
Everyone related to him in a special way; they liked him. He
was always the best. In this case I would say that he is a narcis-
sistic personality.”
Each participant focused on other details in the same case.
Participant 2 focused on the nature of the interpersonal relations
of the patient (case). Participant 9 focused on the accident de-
tails. Participant 6 focused on details describing the patient’s
workplace. This might be one explanation for the differences
between the diagnoses given by the participants.
Results Reg arding Criteria Used to Dec ide on
1) Four categories of the criteria used by the participants to
decide similarities between cases were identified:
a) Diagnosis as it appears in the literature (e.g., “cases 2 and
3 are similar because they are bonarcissists”). Narcissist is a th
Copyright © 2013 SciRes. 3
Copyright © 2013 SciRes.
Table 2.
Diagnosis given to each case by the participants.
Case 1 Case 2 Case 3 Case 4 Case 5 Case 6 Case 7 Case 8 Case 9
personality with
narcissistic traits
personality Borderline
personality Narcissistic
personality Dependent
personality and
obsessive traits
personality and
traits with
personality Schizoid
Narcissistic per-
sonality, obsessive
traits, and de-
pendent traits
personality and
personality with
narcissistic traits
traits and
personality Narcissistic
personality Dependent
personality Narcissistic
personality Compulsive
personality and
traits with
No personality
No personality
disorder Compulsive
personality and
traits with
personality Narcissistic
personality Borderline
personality and
narcissistic traits
personality, and
personality and
personality with
narcissistic traits
personality Borderline
personality and
obsessive traits,
and histrionic
with paranoid
personality and
traits and
personality Narcissistic
personality Narcissistic
personality and
personality and
disorder Narcissistic traits Compulsive
ersonality, and
borderline traits
traits with
No personality
disorder Borderline
personality Borderline
personality Narcissistic
personality Compulsive
Personality Histrionic traits
personality disorder that can be found in the literature.
b) Clusters of disorders (e.g., “cases 2 and 5 are similar be-
cause they present the same cluster, while case 8 belongs to a
different cluste r”).
c) Severity of the personality disorder (e.g., “cases 2 and 4
are more serious cases of personality disorder than case 5”).
d) Personal criteria (e.g. “I don’t l ike either of them …”; “they
are both immature”; “in both cases the problem began after
their marriage”; “in both cases they had the same style of life”).
2) The participants used different criteria to decide on simi-
larities between the same sets of cases.
3) The criteria which psychiatrists used to explain similarities
could be divided into two categories: formal/nonformal criteria
and subtle/general criteria.
a. Formal/nonformal criteria. For the purpose of this study,
formal criteria were those criteria found in the literature that the
subjects used for their decisions on the similarity between cases.
These criteria included personality disorder categories; the
categorization of traits; and rules describing the personality
disorder, the traits and the axis. Examples from the interviews
are as follows:
“They both have a borderline personality disorder, while
the other case is narcissistic.”
“In this pair, axis 1 is dominant, while in the other case
axis 2 is dominant.”
“There is meaning to the relationship between the patient
and her psychiatrist.”
Table 3.
Classes of differences be t w een diagnoses.
Classes of d ifferences
between diagnoses
Identification of personality
disorders In four cases (45%), at le ast two participants did not identify a personal ity disorder.
Different diagnoses of
personality disorders
Psychiatrists diagnosed differently the same cases. Examples:
1) Case 3 was diagnosed by three participants (33%) as narcissistic, whil e three other participants id entified the case
as borderline personality d isorder.
2) Diagnoses for case 6:
a) Four participa nts did not identify any personality dis order.
b) One participant identified dependent traits.
c) Two participants identified dependent personality.
d) One participant identified schizoid personality disorder.
e) One participant identified narcissistic personality disorder.
Severity of personality
disorder In some cases there were differences in the severity attributed to the disorder. While some particip ants identify
personality disorders in a case, others identified only personality disorder traits.
Number of personality disorders
attributed to each patient
For instance, in case 7, four participants (45%) identifie d only narcissistic personality disorder. One participant
identified narc issistic personality disorder with obsessive-compul sive disorder. Three participants identified
narcissistic, dependent, and obsessive-compulsi v e personal ity disorders.
Nonformal criteria were those that were not found in the lit-
erature or occurred when personal meaning was given to formal
criteria. Examples from the interviews are:
“This is his first admission in the hospital.”
“They didn’t achieve anything in life.”
“Only her economic status is important for her.”
“Problems began after marriage.”
“He is cold. Doesn’t express feelings.”
“They both make me nervous. I don’t like them.”
From these results we can conclude that psychiatrists use
both formal and nonformal criteria to determine similarity be-
tween cases. In some cases the nonformal criteria had more
weight in making the similarity decisions, while in other cases
it carried less weight. The following examples will clarify this
Example 1: When participant 6 was provided with cases 2, 5,
and 8, he identified cases 5 and 8 as similar because “both have
a gray personality and style, while case 2 is much more color-
ful.” However, his main diagnosis for both cases 2 and 5 was
narcissistic personality disorder, while he diagnosed case 8 as
obsessive-compulsive disorder. Use of DSM-IV diagnostic cri-
teria would lead to the decision that cases 2 and 5 are similar.
Example 2: When participant 3 was provided with cases 2, 3,
and 5, she identified cases 3 and 5 as similar because “both
have a need for attention. Case 2 also needs attention, but he is
colder; he doesn’t express affection”. The basis for similarity
was nonformal, by an idiosyncratic criterion. Moreover, she
diagnosed case 2 as borderline and narcissistic personality dis-
orders, case 3 as borderline and dependent personality disorders,
and case 5 as narcissistic. Therefore, if the DSM-IV criteria had
been the basis for her similarity decisions, it would be unlikely
that she would have chosen cases 3 and 5, to whom she in fact
gave different diagnoses. It would be more reasonable to de-
termine cases 2 and 3 as similar, since they both suffered from
borderline personality disorder. This demonstrates that idiosyn-
cratic/nonformal criteria were more heavily weighted in this
b. General/subtle criteria. The results indicated that the par-
ticipants used both general and subtle criteria. General criteria
are those that require a rough level of diagnosis. For instance,
personality disorder is a general criterion, whereas one specific
rule appearing in the DSM-IV under that personality disorder
category is a more subtle criterion. Examples from the inter-
views are:
General criteria:
“They are both borderline, while the other case is narcis-
“They are both from the same age range.”
“They are both women.”
Subtle criteria:
“He doesn’t need to be in relationships with others.”
“She hurt herself, feels emptiness.”
“There are extreme fluctuations in their mood.”
Another example of using subtle criteria was when the same
diagnoses were given to three cases, but one was determined to
be dissimilar to the other two, e.g., because “she is more sta-
ble”. Here the severity attributed to the case required subtle
We therefore found no preference for the use of general over
subtle criteria.
4) It was found that heuristics impacted on the criteria by
which the participants determined similarities.
Heuristics are mental shortcuts, “rules of thumb”, which do
not guarantee a solution to the problem, but more often than not
do succeed and save a lot of time and effort in the process (Ey-
senck & Keane, 1995). Heuristics, which are commonly used in
decision making, may lead to faulty reasoning or conclusions
(Elstein, 1999). Representativeness and availability are exam-
ples of such heuristics (Kahneman, Slovic, & Tversky, 1982).
The impact of availability heuristic
There are situations in which people are prone to overesti-
mate the frequencies of easily recalled events (Elstein, 1999;
Kahneman, Slovic, & Tversky, 1982). This judgmental heuris-
tic is calle d av ailability . E xa mpl e from t h e interviews:
While dealing with case 2, participant 9 interpreted the de-
tails of that case and made the diagnosis in a very unique way.
She was the only participant who diagnosed this case as psy-
chotic, and she argued that he did not suffer from a personality
disorder. Here are some justifications that she gave:
Copyright © 2013 SciRes. 5
“He has thoughts that he is a prophet, that he has fire in
his bones. He always murmurs. He believes that if he does
something, he will prevent things from happening. His
ideas are not coherent … It seems that the things in the
interview are fragments, and the only connection between
the different parts of the interview is that they are succes-
sive…It seems as if he is suffering from schizophrenia,
though it is early to decide.”
It was found that a special program dealing with many psy-
chotic cases, in which one of the participants had participated
for eight months prior to this study, had an impact on the diag-
noses she gave and the similarity criteria she used. The cases
the subject had dealt with prior to this study were easily
brought to mind. As a result, her decisions were significantly
based on these cases.
The impact of the representativeness heuristic
When deciding on the probability that object A belongs to
class B, people typically rely on the “representativeness” heu-
ristic, in which probabilities are evaluated by the degree to
which A is representative of B, that is, by the degree to which
A resembles B (Kahneman, Slovic, & Tversky, 1982).
For example, when participant 3 was provided with cases 3,
5, and 7, she identified cases 5 and 7 as similar because “…
nurse and physician [case 5 was a physician and case 7 was a
nurse], this combination is achieving, persistence in work, high
functional lev el, consistent …”
Participant 3 had a mental picture of people who work in
medical occupations. Therefore, it might have been that she felt
a high probability that cases 5 and 7 belonged to this group.
This heuristic therefore influenced her decisions on similarity.
5) Similarity process involves the search for local similari-
While deciding on similarities between cases, psychiatrists
focus only on part of the details of each case. The following
example clarifies this point:
When provided with the cases 3, 5, and 7, participant 6 de-
cided that cases “5 and 7 are similar because they are both
narcissistic. There is something obsessive in both cases. The re
is a failure in relating to their kids”. When provided with the
cases 5, 7, and 9, the participant determined that 7 and 9 are
similar because they are both “extravagant, dramatic, and deal-
ing with emotions” (as opposed to 5 and 7). In another situation
the participant noted “case 9 is similar to cases 2 and 4 because
in all those cases there are changes in mood, which are ex-
pressed in their willingness to die. In both cases they harm
themselves”. According to this participant those are not charac-
teristics of case 7. For each comparison the participant focused
on different characteristics of each case.
6) In 85% of the three-case sets, more than half of the sub-
jects chose the same two cases as similar. Moreove r, in 65% of
the case sets, two thirds of the participants chose the same pair
of cases. However, the explanations given to make these
choices tended to be different, i.e., subjects tended to identify
the same cases as similar, but gave different explanations.
Use of Idiosyncratic Personal Knowledge
This research aimed at identifying the criteria by which psy-
chiatrists decide similarities between personality disorders.
Findings indicate that criteria consisted of nonformal and per-
sonal knowledge. Moreover, sometimes these nonformal/per-
sonal criteria contradicted the diagnoses they identified. This
result lends support to the important role of personal knowledge
in reasoning which challenge the RBR models. RBR models
emphasize on the importance of abstract information in prob-
lem solving and ignore the value of knowledge of a specific
event and specific experiences (Cohen, 1996). The personal
knowledge point of view, on the other hand, sees in knowledge
of specific episodes a key to successful problem solving. Per-
sonal knowledge is defined as the unique frame of reference
and knowledge of self, is central to the individual’s sense of
self (Higgs & Titchen, 1995), and is a result of the individual’s
personal experiences (Butt, Raymond, & Yamaguishi, 1982).
Much of the knowledge a practitioner uses in problem solving
and making clinical judgments is tacit and individual (Polanyi,
1958; Carroll, 1988). In recent years, there has been increasing
concern about the growing gap between research-based knowl-
edge taught in professional schools and the practical knowledge
and actual competencies required of practitioners in the field
(Schon, 1987). The author argues that in order to deal with the
crisis in professional knowledge, we need to recognize that
outstanding practitioners do not have more professional
knowledge, but more “wisdom”, “talent”, “intuition”, or “art-
istry”. By finding those personal criteria by which the psychia-
trists in this study identified similarities may contribute to de-
creasing the gap that Schon warn us from.
Local-Based Similarities Decisions
Other result of this study is that psychiatrists’ decisions re-
garding similarities between cases were based only on part of
the details of each case. This result was also found in their di-
agnoses processes. Psychiatrists focused on only part of the
evidence and did not take into consideration all the information
described in those cases. It is my understanding that they
searched for local coherence rather than a global coherence. An
underlining idea of RBR is that the problem solver, by chaining
together all the relevant domain’s rules, and therefore consider
all the information, is looking for global rather than local co-
Use of Heuristics
It was found that heuristics impacted on the criteria by which
the participants determined similarities. Both representativeness
and availability heuristic are similar-based as it depends on a
memory retrieval process that in turn depends on the similarity
between an encoded event and retrieval cues (Sloman & Rips,
1998). Thus, it is not surprising that it characterize the psychia-
trists who were confronted with a similarity task.
In 85% of the three-case sets, more than half of the subjects
chose the same two cases as similar. Moreover, in 65% of the
case sets, two thirds of the participants chose the same pair of
cases. However, the explanations given to make these choices
tended to be different, i.e., subjects tended to identify the same
cases as similar, but gave different explanations.
Application to Teaching
One problem of today’s teaching is the growing gap between
research-based knowledge and actual competencies required of
practitioners in the field (Eshach & Bitterman, 2002). Psychia-
trists are not always aware of their own personal knowledge
Copyright © 2013 SciRes.
and therefore are not always aware of their own as well as of
other psychiatrists’ criteria. These primary criteria have an
impact on decision-making processes, which in most cases are
based on similarity decisions. In addition, CBR might lead to
faulty decisions. For instance, one might be tempted to use an
old case blindly, relying on previous experience without vali-
dating it in the new situation (Kolodner, 1993). One strategy to
minimize such a cognitive bias is to call these processes to the
attention of the physician and medical educator, in the hope that
when people are aware of these tendencies, they will take steps
to overcome them. Therefore, educators should take such proc-
esses into account when they design learning environments.
The present study provides many examples of criteria used in
the comparison process and thus might be useful as a teaching
aid. The results of this research may assist psychiatrists to be
aware of their own and others’ idiosyncratic c r iteria. Thi s might
also decrease the gap between book knowledge and the per-
sonal/practical experience of physicians, a concern that was
expressed by Schon (1987). In other words, researchers should
make efforts to make the implicit personal/practical experience
of physicians more explicit. Educators, on their turn, should
“use” explicit teaching—which refers to teaching that focuses
on explicit awareness (mediated by verbal interactions) of types
of cognitive procedures (i.e. strategies) being used in specific
instances (Zohar & Peled, 2007; Zohar & Ben David, 2008)—
to teach those, yet, tacit personal cognitive procedures. This
research could be expanded into other domains that rely on
similarity diagnosis.
The author would like to thank the psychiatrists who partici-
pated in the study and Miss Ruth Singer for editing the manu-
American Psychiatric Association (APA) (1994). Diagnostic and statis-
tical manual of mental disorders ( 4th ed.). Washington, DC: APA.
BarOn, E. (1993). Locally coherent views: Toward a unifying theory of
mental capacity and local coherence. Unpublished manuscript.
Barrows, H. S., & Feltovich, P. J. (1987). The clinical reasoning proc-
ess. Medical Education, 21, 86-91.
Boshuizen, H. P. A., & Schmidt, H. G. (1995). The development of
clinical reasoning expertise. In J. Higgs, & M. Jones (Eds.), Clinical
reasoning in the health professions. Oxford: Butterworth-Heinemann
Butt, R., Raymond, D., & Yamaguishi, L. (1982). Autobiographic
praxis: studying the formation of teachers’ knowledge. Journal of
Curriculum Theorizing, 7, 87-164.
Carroll, E. (1988). The role of tacit knowledge in problem solving in
the clinical setting. Nurse Education Today, 8, 140-147.
Cohen, G. (1996). Memory in the real world (2nd ed.). Hove: Psychol-
ogy Press.
Elstein, A. S. (1999) Heuristics and biases: Selected errors in clinical
reasoning. Academic M edicine, 74, 791-794.
Eshach, H., & Bitterman, H. (2002). From case-based reasoning to
problem-bas ed lea rning. Academic Medicine, 78, 491- 496.
Eysenck, M. W., & Keane, T. M. (1995). Cognitive psychology: A
student’s handbook (3rd ed.). Hove: Psychology Press.
Goldstone, R. L. (1994). The role of similarity in categorization: Pro-
viding a groundwork. Cognition, 52, 125-157.
Goodman, N. (1972). Problems and projects. New York: Bobbs-Merrill.
Guba, E. G., & Lincoln, Y. S. (1989). Fourth generation evaluation.
Beverly Hills, CA: Sage Publications.
Higgs, J., & Jones, M. (1995). Clinical reasoning. In J. Higgs, & M.
Jones (Eds.), Clinical reasoning in the health professions. Oxford:
Butterworth Heinemann Ltd.
Higgs, J., & Titchen, A. (1995). Propositional, professional and per-
sonal knowledge in clinical reasoning. In J. Higgs, & M. Jones (Eds.),
Clinical reasoning in the health professions. Oxford: Butterworth-
Heinemann Ltd.
Kahneman, D., Slovic, P., & Tversky, A. (1982). Judgment under un-
certainty: Heuristics and biases. Cambridge: Cambridge University
Press. doi:10.1017/CBO9780511809477
Kaplan, H. I., & Sadock, B. J. (1996). Concise textbook of clinical
psychiatry. Philadelphia, PA: Lippincott, Williams & Wilkins.
Kassirer, J. P., & Kopelman, R. I. (1991). Learning clinical reasoning.
Baltimore, MD: Williams & Wilkins.
Kelly, G. A. (1955a). The psychology of personal constructs (Vol. 1: A
theory of personality). New York: W.W. Norton & Co.
Kelly, G. A. (1955b). The psychology of personal constructs (Vol. 2:
Clinical diagnosis and psychotherapy). New York: W.W. Norton &
Kolodner, J. (1993). Case-based reasoning. San Mateo, CA: Morgan
Kaufmann Publishers.
Kolodny, J. A. (1994) Memory processes in classification learning: An
investigation of amnesic performance in categorization of dot pat-
terns and artistic styles. Psychological Science, 5, 164-169.
Leake, D. B. (1996). CBR in context: The present and future. In D. B.
Leake (Ed.), Case-based reasoning: Experiences, lessons, & future
directions. Menlo Park, CA/Cam bri dge, M A: AA AI Pr ess/M IT Pres s.
Morey, L.C. (1998). Personality diagnosis and personality disorders. In
R. Hogan, J. Johnson, & S. Briggs (Eds.), Handbook of personality
psychology. San Diego, C A: Academic Press.
Mostow, J. (1983). Machine transformation of advice into a heuristic
search procedure. In R. S. Michalski, J. G. Carbonell, & T. M.
Mitchell (Eds.), Machine learning: An artificial intelligence ap-
proach: Vol. 1. Los Altos, CA: Kaufmann. Norman, G. R., Coblentz,
C. L., Brooks, L. R., & Babcock, C. J. (1992). Expertise in visual
diagnosis: A review of the literature. Academic Medicine, 67,
Patton, M. Q. (1990). Qualitative evaluation and research methods
(2nd ed.). Newbury Park, CA: Sage Publications.
Polanyi, M. (1958). Personal knowledge: Towards a post-critical phi-
losophy. London: Routledge & Kegan Paul.
Rissland, E. L., & Skalak, D. B. (1991). CABARET: Rule interpreta-
tion in a hybrid architecture. International Journal of Man-Machine
Studies, 34, 839-887. doi:10.1016/0020-7373(91)90013-W
Rumelhart, D. E. (1989). Toward a microstructural account of human
reasoning. In S. Vosniadou, & A. Ortony (Eds.), Similarity and ana-
logical reasoning. New York: Cambridge University Press.
Schon, D. A. (1987). Educating the reflective practitioner: Toward a
new design for teaching and learning in the professions. San Fran-
cisco, CA: Jossey-Bas s.
Sloman, S. A., & Rips, J. L. (1998) Similarity as an explanatory con-
struct. In A. S. Sloman, & L. J. Rips (Eds.), Similarities and symbols
in human thinking. Cambridge, MA: MIT Press.
Strauss, A. L. (1987). Qualitative analysis for social scientists. New
York: Cambridge University Press.
Vosniadou, S., & Ortony, A. (1989). Similarity and analogical reason-
ing. New York: Cambridge University Press.
Voss, J. F., & Post T. A. (1988). On the solving of ill-structured prob-
lems. In M. T. H . Chi, R. Glaser, & M. J . Farr (Eds.), The nature of
expertise. Hillsdale, NJ: Lawrence Erlbaum Associates.
Williams, S. M. (1992). Putting case-based instruction into context:
Copyright © 2013 SciRes. 7
Copyright © 2013 SciRes.
Examples from legal and medical education. Journal of the Learning
Sciences, 2, 367-427. doi:10.1207/s15327809jls0204_2
Zohar, A., & Peled, B. (2008). The effects of explicit teaching of me-
tastrategic knowledge on low- and high-achieving students. Learning
and Instruction, 18, 337-353. doi:10.1016/j.learninstruc.2007.07.001
Zohar, A., & Ben David, A. (2008). Explicit teaching of meta-strategic
knowledge in authentic classroom situations. Metacognitive Lea r n in g ,
4, 59-82. doi:10.1007/s11409-007-9019-4