The concept of knowledge in KM:

a dimensional model
Bertolt Meyer and Kozo Sugiyama

Abstract
Purpose – This paper seeks to sharpen the concepts of tacit, implicit and explicit knowledge by linking
them to findings from cognitive psychology and memory science and thus finding a possibility for
measuring non-explicit knowledge.
Design/methodology/approach – A review of KM and cognitive science literature leads to a
dimensional model of knowledge types that links the concepts from KM to more specific concepts from
psychology. One central assumption of the model was empirically tested and put into practice in one
Bertolt Meyer is a doctoral small-scale KM project.
student in the Department Findings – The concepts in KM can be linked to concepts from psychology and thus receive theoretical
of Organizational support. The developed model enables psychometric access to a part of non-explicit knowledge
Psychology, Humboldt through structural assessment techniques. Furthermore, the model has proven to be of value in a
University Berlin, Berlin, practical application in KM.
Germany. Kozo Sugiyama Research limitations/implications – The experiment and the practical application are too small in
is a Professor of Knowledge scope to provide full support for the model. Further research is required.
Structure at the Graduate Practical implications – A fraction of non-explicit knowledge can be measured with structural
School of Knowledge assessment techniques. This can be used in the quantitative evaluation of KM projects as these
Science, Japan Advanced techniques allow the quantification of individual knowledge increase. Furthermore, a detailed analysis of
Institute of Science and individual project-relevant knowledge is useful for post-project analysis.
Technology, Ishikawa, Originality/value – The paper integrates findings from several scientific fields for use in KM and
Japan. presents a novel view of classic KM concepts. The developed model is of importance to both
researchers and practitioners.
Keywords Tacit knowledge, Explicit knowledge, Modelling, Knowledge management,
Structural analysis
Paper type Research paper

Introduction
The significance of knowledge as a vital resource for the world’s economies has been
underlined in science and politics (European Council, 2000; Stehr, 1994); it forms the basis
for innovation and economic success (Davenport and Prusak, 1998; Drucker, 1993; Nonaka
and Takeuchi, 1995; Scholl, 2004; Schreyögg and Geiger, 2003). Knowledge management
(KM) is the strategy for creating, accessing and supporting this vital resource. However, the
field of knowledge management is unstructured and scattered, Despres and Chauvel (2000,
p. 57) refer to it as a ‘‘patchwork’’.
The concept of tacit knowledge in particular (Polanyi, 1958, 1966) is credited with a key role
in organizational performance (Nonaka and Konno, 1998; Nonaka and Takeuchi, 1995). It is
at the same time one of the most blurred concepts in management literature (Busch et al.,
2001) and there is an argument whether Polanyi, who coined the concept of tacit knowledge
and Nonaka, who introduced it into knowledge management (Nonaka and Takeuchi, 1995),

DOI 10.1108/13673270710728213 VOL. 11 NO. 1 2007, pp. 17-35, Q Emerald Group Publishing Limited, ISSN 1367-3270 j JOURNAL OF KNOWLEDGE MANAGEMENT j PAGE 17
 ‘‘ Tacit knowledge is difficult to define because there is no
single, broadly accepted definition for the term ‘knowledge.’ ’’

are actually referring to the same thing (Li and Gao, 2003). This uncertainty in regard to
definition of the construct is contradictory to the agreement that ‘‘an increased focus on the
handling of non-explicit knowledge might pose a considerable value-creating factor’’ (FAW,
2001, p. 7, own translation). Therefore further investigation and the development of a sound
conceptual framework is necessary. It is the aim of this paper to link the concepts of
individual implicit, explicit and tacit knowledge with findings from memory, cognition and
knowledge science by developing a two-dimensional model of knowledge types. In this way,
the concepts are not only sharpened but possibilities for their measurement are discussed.
The paper concludes with an experiment that provides some empirical support for the model
and with a brief report on an application of the model in practice.

The concept of knowledge

Tacit knowledge is difficult to define because there is no single, broadly accepted definition
for the term ‘‘knowledge’’. The first philosophical attempt to define knowledge in Plato’s
dialogue of the Theaitetos (Eigler, 1990) described knowledge as ‘‘justified true belief’’. This
introduces truth as a required feature of knowledge (in order to distinguish it from errors).
From the cognitive viewpoint, individual knowledge is simply the content of human long-term
memory (Strube and Schlieder, 1998). One way of bringing the two views together is to
include constructivist aspects. Since constructivism assumes that every individual mentally
constructs their own environment based on their sensory input, there is no such thing as
objectivity or absolute truth, because there is no objective depiction of reality (FAW, 2001).
That is why the term ‘‘viability’’ is introduced. ‘‘Actions, concepts and conceptual operations
are viable if they fit to the intentions or descriptions for which they are used’’ (von Glasersfeld,
1996, p. 43, own translation). This allows the inclusion of an assessment in the concept of
knowledge that does not require objective truth:
Knowledge is not a picture or representation of reality; it is much more a map of those actions that
reality permits. It is a repertoire of concepts, semantic relationships and actions or operations that
have proven to be viable for the attainment of our goals (von Glasersfeld, 1997, p. 202, own
translation).

From this perspective, knowledge contains an assessment in the way that it contains maps
of certain aspects of the world that proved to be viable.

Definition
At this point, we propose a general definition for knowledge that includes an aspect of
assessment and that is applicable to both individual and organizational knowledge. We
base this definition on Güldenberg’s (1999) definition of knowledge as structural connectivity
patterns and state:
Knowledge is defined as a set of structural connectivity patterns. Its contents have proven to be
viable for the achievement of goals.

Based on constructivist assumptions, this definition avoids the term ‘‘representation of

reality’’. It pays tribute to the fact that mental models of an individual are the result of a
construction of environment, which can be very different from one individual to another
(Opwis and Lüer, 1996). The term ‘‘structural connectivity patterns’’ allows the inclusion of
knowledge on different collective levels (individual and organizational), since organizational
knowledge is embedded in the system or structure of the organization. The stress on the fact
that knowledge has proven to be viable underlies the assessment that is a feature of human
knowledge.

j j
PAGE 18 JOURNAL OF KNOWLEDGE MANAGEMENT VOL. 11 NO. 1 2007
 The connection between individual and organizational knowledge
Until now, we have only referred to individual knowledge. However knowledge management
aims at improving both individual and organizational knowledge. Individual knowledge is a
precondition for organizational knowledge which results from the publication of technical
and/or individual knowledge and of its consolidation in organizational communication
structures (Klimecki and Thomae, 2000). This consolidation of individual knowledge in
organizational structures (e.g. in methods, models, documentation and culture) is also
referred to as the organizational knowledge base (Rehäuser and Krcmar, 1996, p. 15).
According to Damerow and Lefèvre (1998), such external representations have the same
psychological functions as internal, individual representations and are based on the same
mental capabilities. This individual knowledge enlarges the organizational knowledge base
(Amelingmeyer, 2004, p. 122 ff.) and individual learning is a central element in organizational
learning (Argyris and Schön, 1999, pp. 20 ff.). All in all, the organizational knowledge base,
which comprises the very entity that KM seeks to optimize, is based on individual
knowledge. Therefore, insights from the field of cognitive science and memory research can
and should be integrated with aspects in knowledge management.

Non-explicit knowledge in KM literature

Individuals can perform actions without being able to explain them and they can explain
actions without being capable of performing them (Dick and Wehner, 2002). From such
observations, Polanyi (1958; 1966) concluded the existence of a silent dimension of
knowledge which cannot be articulated: tacit knowledge. A similar typology is introduced by
Spender (1996), who differentiates between implicit (produced through action) and explicit
(produced through communication) knowledge. Polanyi himself distinguished between
explicit knowledge and tacit knowledge according to the differentiation between Können
(being able to do sth.) and Wissen (knowing) in the German language (Polanyi, 1985, p. 16).
Similarly, articulable knowledge is referred to as explicit knowledge by Nonaka and Takeuchi
(1995), knowledge that is difficult to articulate or cannot be articulated at all is referred to as
tacit knowledge, too. Furthermore, they postulated a mechanism for converting implicit into
explicit knowledge, thus fuelling the boom of knowledge management approaches (Schütt,
2003). However, Li and Gao (2003) argue that Nonaka’s understanding of tacit knowledge
differs from Polanyi’s concept and criticise their synonymous use in literature. The authors
stress that Nonaka and Takeuchi and Polanyi referred to two different observations in two
fundamentally different cultural contexts. Polanyi studied European scientists, whereas
Nonaka and Takeuchi studied factory workers in Japan. Li and Gao state:
It is out of Polanyi’s argumentation for a careful differentiation between tacitness and implicitness,
but from his terminology, tacitness is evidently different from impicitness [sic]. Implicitness, an
other [sic] form of expressing knowing, does exist. It implies that one can articulate it but is
unwilling to do that [. . .]. [. . .] When Nonaka and Takeuchi used Polanyi’s dichotomy [. . .] we can
see that actually what they mean by ‘‘tacitness’’ includes ‘‘implicitness’’ (Li and Gao, 2003, p. 8).

The fact that implicit and tacit knowledge are described as two separate things and the hint
at different levels of codifiability points towards a dimensional character of non-explicit
knowledge (see also Kogut and Zander, 1992). The dimension spans between the poles
explicit knowledge and tacit knowledge. Following Li and Gao, implicit knowledge lies
somewhere in between. Knowledge elements can be classified into this continuum based on
the degree of their codifiability (European Foundation for the Improvement of Living and
Working Conditions, 2004), compare Figure 1.

‘‘ Knowledge management aims at improving both individual

and organizational knowledge. ’’

j j
VOL. 11 NO. 1 2007 JOURNAL OF KNOWLEDGE MANAGEMENT PAGE 19
 Figure 1 Dimensional classification of knowledge

However, the above citation underlines the difficulty in clearly defining the constructs implicit
and tacit knowledge. The following sections will therefore be used to elaborate on
psychological and cognitive findings from the fields of memory research and cognitive
science. This appears appropriate as Li and Gao, Polanyi and memory research refer to
individual knowledge. These findings will then be integrated into the dimensional
classification of knowledge, allowing a clearer definition and specification of non-explicit
knowledge.

Models from memory research

In this section, findings from neuroscience and memory science are outlined that will each
be connected to the concepts employed in KM. For the following descriptions, the definition
for memory employed by Sinz (1979) is used:
The term memory describes the storage that depends on the learning of ontogenetically acquired
information that selectively inserts itself into phylogenetical neuronal structures and can be
recalled at any given point in time, e.g. that can be made available for situationally appropriate
behaviour (Sinz, 1979, quoted in Markowitsch, 2002, own translation).

It can thus be argued that individual knowledge is stored in memory (Strube and Schlieder,
1998). Generally speaking, memory models either describe the structure of memory or the
processes that are active in memory (Tulving, 2002). In the following, two common and
extensive memory models are presented: the content-related memory model (Markowitsch,
1992, 1999, 2002; Squire and Frambach, 1990; Squire et al., 1993; Tulving, 1972, 1995,
2002) and the multimodal theory of memory (Engelkamp, 1991; 1998; Engelkamp and
Pechmann, 1993; Engelkamp and Zimmer, 1994). The latter includes both structure and
processes; the former is a classification approach. After these two models are laid out,
knowledge representation models that build on them are described. Afterwards, the models
are incorporated into a dimensional system of memory types.

Content-related memory model

This memory model is based on neuroanatomical findings by Markowitsch (1992; 1999;
2002), Tulving (1972; 1995; 2002) and Squire and his colleagues (Squire and Frambach,
1990; Squire et al., 1993).

Working memory
Firstly, the model postulates a memory with a short memory span of a few minutes that all
information needs to pass through in order to be permanently stored in the long-term
memory (Markowitsch, 2002, p. 85). It can be understood as that part of memory that is
active at a certain point in time (Markowitsch, 2002, p. 85) and is therefore referred to as
working memory.
Several findings indicate that the working memory is made up of several modality-specific
subsystems, e.g. for verbally and visually coded information, that are coordinated by a
central entity (see Squire et al., 1993 for an overview). The capacity of the verbal working
memory is five (plus/minus two) informational units (chunks) (Markowitsch, 2002); the
capacity of the visual working memory is assumed to be four objects that can have up to 16

j j
PAGE 20 JOURNAL OF KNOWLEDGE MANAGEMENT VOL. 11 NO. 1 2007
memorable features (Vogel et al., 2001). On a neural level, network theories are most popular
for describing memory processes (Markowitsch, 1999, 2002).

Long-term memory
Within long-term memory, where the maximum length of storage is practically unlimited,
several different memory systems can be differentiated according to their content (long
refers to a span beyond a minute, cf. Markowitsch, 1999). The youngest part of memory from
an evolutional biological point of view is episodic memory (Tulving, 2002). ‘‘It consists of
singular events that can be specified according to time and place’’ (Markowitsch, 2002,
p. 88, own translation). Together with semantic memory that stores general facts about the
world, it belongs to the declarative memory system. Episodic memory builds on semantic
memory.
According to Squire et al., ‘‘Declarative Memory is fast, it is not always reliable (i.e. forgetting
and retrieval failure can occur), and it is flexible in the sense that it is accessible to multiple
response systems’’ (Squire et al., 1993, p. 458).
The content-related memory model states that humans also possess reflexive or
non-declarative memory. ‘‘Non-declarative memory is slow [. . .] reliable and inflexible’’
(Squire et al., 1993, p. 458). Reflexive memory is differentiated into three subsystems:
procedural memory, the priming system and the part of memory that is responsible for
conditioning. For the non-declarative memory systems, Squire synonymously employs the
term ‘‘implicit memory’’ (Squire et al., 1993, p. 471). The procedural memory system
contains skills and habits: ‘‘Skills are procedures (motor, perceptual and cognitive) for
operating in the world; habits are dispositions and tendencies that are specific to a set of
stimuli and guide behaviour’’ (Squire et al., 1993, p. 471). Under certain conditions, these
can be acquired unconsciously. It should be noted that procedural memory does also
contain skills that are not on a motor level, but on a perceptive and/or a cognitive level
(Squire et al., 1993, p. 472). Non-declarative memory can be acquired independently of
declarative memory (Squire and Frambach, 1990). Figure 2 illustrates the model.
With reference to knowledge management, findings from experiments in learning artificial
grammars are of special importance. In these experiments, subjects receive lists of
meaningless words and are told that the syntax of these words does in fact follow a set of
rules, which remain undisclosed. After the subjects are presented with ‘‘valid’’ meaningless
words, they are asked to decide whether previously unknown meaningless words obey the
rules or not. Although subjects are unable to explain the grammatical rules on which their
judgement is based, the number of correct decisions is above coincidence (Squire et al.,
1993, p. 473 f.). In these kinds of tasks, it is impossible to determine whether subjects
employ implicit knowledge in terms of procedural knowledge, or whether they employ
incomplete or weak declarative knowledge (Squire et al., 1993, p. 474).

Figure 2 Overview of the content-related memory system

j j
VOL. 11 NO. 1 2007 JOURNAL OF KNOWLEDGE MANAGEMENT PAGE 21
 This statement implies that non-articulable knowledge can have two causes: it is either
procedural or weak declarative. Both have in common that knowledge elements are
accessed subconsciously. For this reason, Markowitsch rejects the synonymous use of
procedural and implicit memory:
Implicit and explicit memory are not two different kinds of memory, they are different forms of
expressing memory or phenomenologically different ways of retrieving specific events or
experiences. Implicit means without making the actual content and its meaning conscious,
explicit means including the associated connotations (time-spatial coordinate structure, the how,
when and where of the encoding process). Explicit recall manifests the recalled information as an
episode that can be personally experienced. The neural structural combinations that are
responsible for implicit and explicit memory processing do differ (Markowitsch, 1999, p. 25, own
translation).

Thus implicit memory describes an unconscious processing of memory contents, whereas

the term explicit memory refers to a conscious mode of processing. Kluwe (2004) arrives at
the same conclusion when he describes implicit knowledge as ‘‘superior performance in
cognitive tasks based on an unconscious use of previously perceived and not intentionally
stored information’’ (Kluwe, 2004, p. 5, own translation). In an analogous way, Kluwe defines
explicit knowledge as conscious recall of previously encoded information.
In addition to conscious and unconscious use of knowledge, there exists the phenomenon
that previously acquired knowledge is not used at all. This so-called ‘‘inert knowledge’’
(Renkl, 1996; Whitehead, 1929) is used to explain the discrepancy between knowledge and
behaviour in pedagogy.
It becomes evident that the concepts employed in KM are not contradictory to concepts in
memory and cognitive science, but for a complete overview, a model needs to be presented
that spans not only different types of memory but also different memory processes. The
multimodal memory model (Engelkamp, 1991, 1998; Engelkamp and Pechmann, 1993;
Engelkamp and Zimmer, 1994) is such a model and is introduced in the next section.

The multimodal memory model

This model includes both process and structural assumptions. In accordance with Tulving’s
differentiation between semantic and episodic memory, Engelkamp and colleagues
introduce a multimodal memory model for episodic memory processes.
The actual memory model is based on the assumption of the existence of two orthogonal
dimensions: sensory-motor and verbal-nonverbal. Within these two dimensions, the authors
postulate a conceptual system linked to modality-specific entry and output systems
(Engelkamp, 1998, p. 35), the so-called sensory-motor systems. Throughout interaction
between the systems, information is represented on two different levels: the conceptual
system operates independently from the modality of the input; on the sensory motor level,
encoding is specific for the modality of input and output (see Figure 3).
Sensory motor systems are differentiated into sensor and motor systems. For simple items
such as single concepts or actions, the authors assume a strict separation of memory
content into different subsystems. Complex material is based on several modalities and thus
on several subsystems. Sensory motor knowledge preserves experiences close to
perception and behaviour. Referent knowledge combines concepts into propositions
(compare next section).
One speciality of Engelkamp’s model is the assumed lack of connection between verbal
sensory motor system and nonverbal sensory motor systems. The authors assume that an
access from the verbal sensory motor system to the nonverbal sensory motor system
requires participation of the conceptual system (Engelkamp, 1991, p. 8).
In order to make an action verbally explicit, a reference to a motor program must be present
in the conceptual system that must be connected to a word node in the verbal sensory-motor
system. Only if this word node is connected to a modality-specific verbal program, can
verbalization take place. Explicit knowledge about acting requires a connection of all three

j j
PAGE 22 JOURNAL OF KNOWLEDGE MANAGEMENT VOL. 11 NO. 1 2007
Figure 3 General architecture of the multimodal memory model

systems with regard to a specific content. If someone is capable of performing an action

without being able to verbalize it, this can have two reasons.
The sensory nodes, e.g. visual nodes, are directly connected to motor programs in the
nonverbal sensory motor system. Such content that is related to a single subsystem can only
occur for simple stimuli and actions (see above). An example would be turning the head
towards a face we recognize in a crowd. The face is the visual sensory node which was
activated by perceiving the face; it is directly connected to the motor program for turning the
head. Knowledge regarding face recognition is accessed without any use of the conceptual
system and is thus not codifiable. This type of knowledge can be labelled embodied
knowledge.

The performance of complex acquired tasks that are difficult to verbalize includes both the
nonverbal sensory-motor system and connected referent knowledge within the conceptual
system. Conceptual knowledge does not include knowledge of words, or known words are
not connected to the word nodes in the verbal sensory-motor system. The connection
between the conceptual system and the verbal system, which was established during
learning, may have faded over time since the verbal sensory-motor system is no longer
required after learning in order to perform the action. Actual performance of the action
requires only the conceptual and the nonverbal sensory-motor system.

Due to the inclusion of actions and behaviour in the memory model, the multimodal memory
model is capable of explaining differing levels of verbalization of behaviour that can be
observed. Non-explicit knowledge acquired over time that was compiled into automated

j j
VOL. 11 NO. 1 2007 JOURNAL OF KNOWLEDGE MANAGEMENT PAGE 23
 actions, such as the expert mastery of a musical instrument, can be explained as a
disassociation of the verbal sensory-motor system for that particular action or concept.
All in all, the models based on Tulving and Markowitsch and the model based on Engelkamp
and Zimmer can explain the existence of memory content which is not consciously
accessible. In Tulving and Markowitsch’s model, this can either be procedural knowledge or
unconscious access to (weak) declarative knowledge. In Engelkamp and Zimmer’s model,
sensory-motor systems without connection to the conceptual system or to the verbal
sensory-motor system are active.
The phenomenon of non-explicit knowledge from the realm of KM thus finds its
correspondence in memory psychology. However, the cognitive dimension of procedural
knowledge and the reference to semantic network structures in previous sections do require
the introduction of higher-level concepts of representation. These will be outlined in the
following section, prior to introducing the model.

Cognitive models for knowledge representation

Until now, different memory structures and their relationships have been described.
Knowledge organization goes beyond this level; it deals with how semantic structures and
productions are actually organized.
Two forms of knowledge representations are propositional representation systems and
rule-based representation systems (Opwis and Lüer, 1996). They will be briefly described in
this section.

Propositional representation systems

Propositional representation systems represent ‘‘verbally articulable information with the
help of special symbol structures, so-called propositions’’ (Opwis and Lüer, 1996, p. 349,
own translation). Two famous propositional systems are semantic networks and cognitive
schemes. Semantic networks are formally depicted as graphs in which nodes represent
linguistic units and edges represent linguistic relations. This approach is primarily based on
Quillian (1968), who assumed a networked organization of individual semantic knowledge. A
problem with semantic networks is their limited expressiveness and the fact that they do not
include methods for dealing with objects in memory. This criticism can be met with an
advancement of semantic networks: cognitive schemata. They refer to a heterogeneous
group of pre-structured representational formats. The two most popular types of cognitive
schemata are frames and scripts (Strube and Schlieder, 1998).
Frames are data structures in which experiences are generalized and that represent
circumstances and expected coherences from a certain realm of reality (Schnotz, 1994).
These representations contain constants and vacancies that store probabilities for other
schemata that can be inserted. In this way, a schema is an instantiable class of a situation. A
proposition is a structure that is created on instantiation of a schema.
A script is a frame for a situation involving several actions, much like a film script for standard
situations. The most famous example is the script for a restaurant visit, in which certain
behaviour such as waiting to be seated, being seated, receiving the menu, ordering, eating,
paying and leaving are organized in a sequential manner. Scripts allow economic
information processing that is steered by expectations.

Rule-based representational models

This form of knowledge representation assumes concurring processes within a production
system. Contrary to cognitive schemata, production systems claim separate storages for
declarative and procedural knowledge. A production rule or production connects a
condition to an action. Declarative knowledge consists of data structures processed in
working memory. Processing takes place by applying production rules to the content of the
working memory (Schnotz, 1994). Declarative knowledge is represented as a semantic
network with edges and nodes and is stored in the declarative long-term memory. The nodes
of the network are knowledge units, the edges between them correspond to certain relations

j j
PAGE 24 JOURNAL OF KNOWLEDGE MANAGEMENT VOL. 11 NO. 1 2007
‘‘ Nonaka and Takeuchi and Polanyi referred to two different
observations in two fundamentally different cultural
contexts. ’’

between these units (Schnotz, 1994, p. 96). Declarative knowledge elements are associated
through productions containing them. This leads to the assumption that knowledge of many
actions or differing patterns of behaviour leads to a higher degree of connectedness
between declarative knowledge elements.
All in all, propositional representation systems and rule-based representational models
describe individual knowledge representations and knowledge processes on a higher level
than the models in the previous section. Note that a network organization of knowledge is
central to all models that have been described in this paper so far.

Structural Knowledge
Since network organization is a central characteristic underlying the models discussed, its
organization is introduced as an independent characteristic of knowledge: Structural
Knowledge. It is ‘‘[. . .] the knowledge of how concepts are interrelated’’ (Jonassen et al.,
1993, p. 4). It is a ‘‘[. . .] hypothetical construct referring to the organization of the
relationships of concepts in long-term memory’’ (Shavelson, 1972, pp. 226-7, quoted in
Jonassen et al., 1993). The authors further state:
Structural Knowledge is also known as cognitive structure, the pattern of relationships among
concepts in memory (Preece, 1976) [. . .]. Structural knowledge has also been referred to as
internal connectedness, integrative understanding or conceptual knowledge (Jonassen et al.,
1993, pp. 4 f.).

This conceptual structure facilitates between declarative and procedural knowledge and
thus conditions the acquisition of procedural knowledge. According to Jonassen et al,
blanks within cognitive schemata are references to other schemata. In this way, the
interrelations between cognitive schemata can be seen as a semantic network with
schemata as nodes. This view is consistent with Quillian’s concept of semantic memory.
Jonassen et al. assume that structural knowledge is always explicit, i.e. the connections
between concepts can always be expressed. However, there are empirical findings
indicating structural knowledge can be non-explicit. Rothe and Warning (1991) tried to
elicitate the structural knowledge of experts in a limited specified knowledge domain
through the structure-laying technique (Scheele and Groeben, 1984). It turned out that the
number of nodes and their labels were similar among subjects while the labelling of the
edges with Klix’ standard semantic relations (Klix, 1984) differed to a great extent. Rothe and
Warning concluded that their subjects generally had substantial difficulties in naming the
edges between knowledge nodes. This leads to the assumption that access to structural
knowledge can be implicit in Tulving’s sense, i.e. present but not consciously accessed. This
assumption is supported by Davis et al. (2003) who assume that the elicitation of structural
knowledge (structural assessment) also captures non-explicit knowledge. The authors state
that tacit knowledge is comprised of the subtle interrelations between concepts and
explicitly indicate the possibility of measuring at least a part of tacit knowledge by structural
assessment. Lee et al. (2002) follow this approach by attempting to capture the
organizational members’ tacit knowledge through knowledge structure elicitation
techniques.
To sum up, the connections in semantic memory can be interpreted as an independent type
of knowledge (structural knowledge). They can be accessed either consciously or

j j
VOL. 11 NO. 1 2007 JOURNAL OF KNOWLEDGE MANAGEMENT PAGE 25
 unconsciously and can thus be non-explicit knowledge (see above). Structural knowledge
can be elicited through several different methods (see Jonassen et al. for an overview).

Integration into a dimensional model

After all relevant terms have been introduced, we return to the concepts from KM in order to
connect them with the concepts outlined. The dimensional figure of knowledge codifiability
presented earlier is replaced by a more detailed model by the end of this section, which is
based on Schindler (2002) who states that the transition between explicit and non-explicit
knowledge is fluid. In his model, the right pole of the dimension is assigned to
non-articulable tacit knowledge that includes capabilities such as maintaining balance and
face recognition. This corresponds to purely sensory-motor memory contents in Engelkamp
and Zimmer’s model (see above). Since this knowledge is rooted into fundamental
neurological mechanisms and is inherited, it is therefore beyond the scope of KM. Schindler
takes the next two sections of his dimension from Nonaka and Takeuchi (1995):
[. . .] tacit knowledge can be segmented into two dimensions. The first is the technical dimension,
which encompasses the kind of informal and hard-to-pin-down skills or crafts captured in the term
‘‘know-how’’. [. . .] At the same time tacit knowledge contains an important cognitive dimension. It
consists of schemata, mental models, beliefs, and perceptions so ingrained that we take them for
granted (Nonaka and Takeuchi, 1995, p. 8).

Thus the technical dimension corresponds to procedural knowledge in the non-reflexive

memory system in Markowitsch’s model (compare Figure 2). It is acquired through motor
skill learning for which well-established theories exist (Fitts and Posner, 1979). The third part
of Schindler’s dimensional model of non-explicit knowledge model corresponds to the
cognitive dimension of tacit knowledge (compare Figure 4).

Extended dimensional model

As noted earlier, from the psychological viewpoint, implicit and explicit memory are different
ways of using memory content: either consciously or unconsciously. At the same time, it was
stated that knowledge can be articulated in varying degrees (see above). Both features can
be seen as dimensions that span an area, onto which the different knowledge types can be
mapped. This knowledge map also allows the inclusion of Nonaka and Polanyi’s concepts
that span several knowledge types.
With reference to the different types of knowledge introduced in previous sections, Polanyi’s
tacit knowledge can be equated with embodied knowledge and procedural knowledge.
Polanyi made no reference to conscious or unconscious use; he only referred to knowledge
that cannot be articulated.
Li and Gao stress that Nonaka’s concept of tacit knowledge extends beyond Polanyi’s view
(see above). We thus assume that it is Nonaka’s cognitive part of tacit knowledge that
surpasses Polanyi and that it does include all unconscious uses of memory, independent of
the fact that they could possibly be verbalized. In this way, unconscious access to structural

Figure 4 Dimensional knowledge classification according to Schindler (2002)

j j
PAGE 26 JOURNAL OF KNOWLEDGE MANAGEMENT VOL. 11 NO. 1 2007
 knowledge and weak declarative knowledge would be mapped to Nonaka’s concept of tacit
knowledge, but not to that of Polanyi. This assumption is supported by the fact that Nonaka
and Takeuchi do assume that non-explicit knowledge elements can be made explicit
through appropriate techniques.
According to Nonaka and Takeuchi, explicit knowledge can always be verbalized. Since
verbalization requires conscious access to memory contents, declarative knowledge (both
semantic and episodic) and the conscious use of structural knowledge can be connected
with this concept. The model is summarized in Figure 5.
It turns out that constructs from KM literature are not contradictory to findings from memory
research and other fields of psychological research. They integrate several constructs in a
way that does seem suitable in practice. Both Nonaka and Takeuchi’s constructs can be
linked to an empirically founded basis.

Empirical support
The potential of the developed framework lies in the fact that it makes a part of non-explicit
knowledge, the unconscious access to structural knowledge, available to psychometric
assessment. There are a number of empirical studies that try to access, elicitate and
measure individual structural knowledge (Davis et al., 2003; Eckert, 1998a, b; Goldsmith
and Johnson, 1990; Goldsmith et al., 1991; Jonassen et al., 1993; Lee et al., 2002;
Schvaneveldt, 1990; Schvaneveldt et al., 1985a, b). The method that is used in all of these
studies is the structural assessment technique (SA, compare Jonassen et al., 1993 for a
detailed description). The result of SA is a graph representing the semantic structure of
individual knowledge within a specified domain. The nodes are knowledge elements, for
example important actions for performing a specific task. The edges represent their relation
from the subject’s viewpoint.

In order to find empirical support for the model’s essential assumption that there exists a
difference between conscious and unconscious access to structural knowledge, an
experiment was conducted.

Figure 5 Extended dimensional model of knowledge types

j j
VOL. 11 NO. 1 2007 JOURNAL OF KNOWLEDGE MANAGEMENT PAGE 27
 Hypothesis and method
If structural knowledge can be accessed both consciously and unconsciously, we assume
that the quantity of structural knowledge an individual can access consciously and
subconsciously is larger than the quantity of structural knowledge that is only available
consciously (that is, the subject can name the relation between two knowledge objects).
Therefore, we put forward the following hypothesis:
A graph that represents individual knowledge in a certain field that only includes labelled
edges (explicit structural knowledge) is smaller (i.e. contains less edges) than a knowledge
graph that contains both edges that cannot be labelled and edges that can be labelled by
the subject.
We used a two-staged, computer-based process to test this hypothesis. Subjects first had to
associate concepts that, to their knowledge, relate to a certain knowledge domain. These
associated words were stored in the computer’s memory and were then, in a second stage,
presented to the subjects as pairs. Subjects had to rate the degree of their relatedness,
usually on a five-point scale. The concept x concept matrix of the paired ratings is the raw
graph of the knowledge organization for the specified knowledge domain. The measurement
error is met by applying the path-finder algorithm on this graph that removes all edges that
violate the triangulation criterion (for details on this procedure and its theoretical
background, compare Schvaneveldt, 1990). These pathfinder-adjusted graphs are
referred to as PFNETs in the literature. PFNETs have proven to be an effective method for
differentiating between experts and novices and are recognized as a valid representation of
individual structural knowledge (Goldsmith and Johnson, 1990; Goldsmith et al., 1991).
PFNETs can be analyzed with methods from graph theory and thus can be represented in a
quantitative way.

In our setting, the second rating task of the test came in two different modes. In one mode,
subjects were only allowed to make a connection between two concepts if they were able to
explicitly state the nature of the connection. This mode was labelled ‘‘explicit’’ mode. In the
other mode, subjects were asked to make quick judgements on the relation between two
concepts without having to explicitly state the nature of the connection. This mode was
labelled ‘‘implicit’’ mode. If the aforementioned hypothesis would hold true, subjects in the
‘‘implicit’’ mode should have significantly more edges in their PFNETS as these would
include both edges that could potentially be labelled and edges that could not. The two
groups were compared using independent samples T-test as the obtained data is metric.

Subjects
In total, 30 graduate students at the department of organizational and social psychology of
Humboldt University Berlin participated in the experiment for an extra seminar credit. All of
them had previously completed a graduate seminar on organizational knowledge
management. The knowledge domain in which subjects had to associate was thus
chosen to be ‘‘knowledge management’’ as all students were expected to have obtained
both declarative and non-declarative knowledge in the field. Assignment to the ‘‘implicit’’
and ‘‘explicit’’ group was random. All subjects completed the test and associated concepts
and edge labels were reviewed by the seminar instructor for face validity.

Results
The results with reference to average individual PFNET degree (number of nodes) and
average size (number of edges) are given in Table I.
Table I shows that subjects in the two groups did not associate a significantly different number
of concepts in the first part of the test. Therefore a significant difference in the number of edges
is not conditioned by the degree of the PFNETs. The average number of edges within the
PFNETs does in fact differ significantly between the groups and, with an effect size of almost
0.80, to a large extent. If the number of edges is adjusted to the number of nodes in subjects’
PFNETs, this effect is even stronger and reaches an effect size of almost 1.

j j
PAGE 28 JOURNAL OF KNOWLEDGE MANAGEMENT VOL. 11 NO. 1 2007
Table I Comparison between explicit and implicit group (T-test for independent samples)
Test mode
Implicit Explicit
(n ¼ 15) (n ¼ 15)
Item M SD M SD p(t) du a

Graph degree (# of associated concepts) 13.40 5.99 10.09 5.66 0.244ns 0.55
Graph size (# of edges) 53.47 45.35 25.53 15.72 0.012* 0.79
Edges per node 3.47 1.40 2.21 1.01 0.037* 0.99

Note: a Effect size Cohen’s d with bias correction for small samples

Discussion
The above results cannot be seen as comprehensive empirical support for the model as the
experiment was too small in scope. However its results support the posited hypothesis and
thus these findings are not contrary to the statements of the model. There seems to be a
difference between explicit structural knowledge and non-explicit structural knowledge in that
subjects see more relations between concepts than they can label. These results correspond
to the results of Davis et al. (2003), who showed that within the multiple correlation of structural
knowledge and declarative knowledge onto a performance-based outside criterion, structural
knowledge shows a significant additional explanation of variance. The fact that declarative
knowledge and structural knowledge together do not fully explain the variance of a
performance-based outside criterion underlines the fact that unconsciously used structural
knowledge only captures a fraction of individual non-explicit knowledge.

Practical application
The main purpose of the model outlined earlier is the possible psychometric access to parts
of non-declarative knowledge. A practical application could thus lie in the use of structural
assessment techniques for evaluating KM approaches. However, another application of the
model is possible and will be outlined in this section.
Apart from inert knowledge, concrete individual knowledge can be specified for each
knowledge type that is specified in the model. It is possible to state what kind of specific
declarative knowledge and skills an individual contributed to a project and what kind of weak
declarative knowledge manifested itself during the course of a project. If structural
knowledge is seen as mediating between declarative and procedural knowledge (see
above), it is also possible to draw references to individual structural knowledge from the
skilful and articulable application of a capability.
Breaking down project-relevant knowledge of individuals into these categories can add
valuable detail to the analysis of project outcomes. At the Japan Advanced Institute of
Science and Technology (JAIST), a framework for process analysis of organizational
knowledge creation in academic research projects is currently being developed (Sugiyama,
2005). Its intention is to provide tools for precise analysis of small projects in order to
discover how project members share contexts, cooperate, obtain, and exchange
knowledge. The use of this framework will enable the identification of factors that are
essential to a project’s outcome.
The framework includes concrete small-scale applications of several Knowledge Creation
(KC) theories such as the Equivalent Transformation Theory (Ichikawa, 1970), and the SECI
Model (Nonaka and Takeuchi, 1995). The framework also includes an earlier version of the
model from Figure 5. The model is used to guide the project leader in post-project evaluation in
specifically stating the different knowledge types for each project member that contributed to
the project’s results. The work guide was developed during the evaluation of a project on
abstraction and media conversion. During this project, team members developed strategies
to formalize puzzles into abstract models (graphs) which could then be implemented in media
that differ from the medium of the original puzzle, e.g. into graphs, blocks, sounds, or robots as
shown in Figure 6 (Maeda et al., 2002; Sugiyama et al., 2003, 2005).

j j
VOL. 11 NO. 1 2007 JOURNAL OF KNOWLEDGE MANAGEMENT PAGE 29
 Figure 6 Project overview

Following project completion, the supervisor, based on his knowledge of the progress of the
project, intuitively identified an organizational and knowledge network among related
persons (see Figure 7) and assembled a list of critical individual knowledge for each
member of the research team based on the dimensional model (compare Table II).
The empty cells in the table indicate that the project supervisor assumes that these
individual knowledge types did not contribute substantially to the project’s outcome.
The use of such fine granular analysis is threefold. First, the project supervisor realizes that it
is more than knowledge of facts that contributes to the outcome of a project. Second,
combined with cross-tables for inter-individual knowledge exchange that is based on the
analysis of knowledge flows (compare Figure 7), knowledge flows within the project can be
analyzed at a deeper level. Third, this analysis can reveal potential areas for future
improvements. For example if a project fails and the post-project analysis reveals a lack of
individual non-explicit knowledge use or exchange, this could be an indicator for areas of
improvement for future projects.

Figure 7 Individual knowledge exchange over different organizations during the project

j j
PAGE 30 JOURNAL OF KNOWLEDGE MANAGEMENT VOL. 11 NO. 1 2007
Table II Individual project-relevant knowledge in a small-scale research project as seen from the project’s supervisor point
of view after the project’s completion
Explicit knowledge Non-explicit knowledge
Unconscious
Conscious access access to Acquired
Declarative to structural Weak declarative structural skills/procedural Embodied
Person knowledge knowledge knowledge knowledge knowledge knowledge

A Spring algorithm How to utilize Logic of puzzles Equivalent Programming Artistic senses
spring algorithm transformation skills for
thinking developing
generators
B Graph drawing Mathematical Logic of puzzles Systems Mathematical System thinking
algorithm, formalizations integration and derivations
Geometry, Graph analysis
theory
C Tutte algorithm Programming
skills
D AIBO control Integration of IT
tools,
programming

Source: Sugiyama (2005)

Sugiyama (2005) concludes that the work flow model which is in part based on this model
was successfully employed in the analysis of a scientific research project and that he
recognized evidence of the concepts. It is also stated that further research in the form of
analysis of further cases is desirable. However in this concrete application, the model has
proven to be of practical use in KM practice in addition to structural assessment.

Conclusion and outlook

In this paper, we have introduced memory models for individual knowledge representation
and linked them to KM concepts such as Polanyi and Nonaka’s concepts of tacit knowledge.
In this way, we arrived at a dimensional model of knowledge types that proved to be of
concrete use for application in a KM project. Some empirical support for the model’s
assumptions was also found.
Future studies that capture structural knowledge in a valid way and are able to demonstrate
an additional explication of variance of a performance-based outside criterion through the
inclusion of structural knowledge will have to be conducted. These will show whether the
proposed dimensional conceptual framework for knowledge receives further empirical
support. This includes the empirical evaluation of the hypothesis that access to structural
knowledge can be unconscious.
As indicated above, current empirical and practical support are still very limited. The
experiment’s small sample size and its homogenous structure (30 HU students) do not allow
generalization of findings. Further experiments that aim at replicating the findings outside
the academic field using larger samples are currently under way, as well as validation
experiments that deal with the test’s predictive validity. At the same time, the outlined
computer-based structural knowledge elicitation test is turned into a modular web-based
application. If the validation experiments prove to be successful, the described test will be
published under the name AST* (Association Structure Test) and will be available for
organizational use. It will then allow an exact quantification of individual knowledge increase
over time through pre-post-analyses. In this way, the test can be used for assessment of
organizational knowledge management initiatives. If they are successful, employees should
have more structural knowledge after the initiative than before. This assumption could be
tested with the AST*.

j j
VOL. 11 NO. 1 2007 JOURNAL OF KNOWLEDGE MANAGEMENT PAGE 31
 Because of the small project size and possible social and cultural issues, the preliminary
application of the model at JAIST is also difficult to generalize. Possible cultural issues
include a higher demand for conformity and a strong awareness for hierarchy in Japanese
project teams. In the presented case, the project leader intuitively assembled the list of team
members’ relevant contributed knowledge and project members may have refrained from
correcting his assessment.
In order to tackle the limitations of the empirical findings, new experiments are being
conducted at Humboldt University Berlin, focusing on the predictive validity of structural
knowledge elicitation and on measuring knowledge increase over time. Therefore, attempts
at predicting students’ grades in future exams by features of their structural knowledge are
being made. Furthermore, the changes of individual implicit and explicit knowledge graphs
after face-to-face interaction are measured with the AST*. In this way, insights on constraints
and enablers of successful knowledge sharing are gained and our understanding of the
importance of shared educational backgrounds is broadened. In further studies, we seek to
find empirical answers to questions such as: ‘‘Do similar or dissimilar educational
backgrounds lead to higher knowledge increases through interaction?’’.
Despite the challenges that lie ahead in the course of empirically supporting the dimensional
model, it has the advantage that knowledge management activities targeting non-explicit
knowledge can be narrowed in their target focus, e.g. with the aim of targeting either motor
skills or cognitive components. However the actual promise of the outlined dimensional
framework lies in its inherent possibility to empirically evaluate knowledge-based activities
that target individual non-explicit knowledge through structural assessment, as an increase
in non-explicit knowledge should also lead to an increase in structural knowledge. This
possibility justifies further investigations based on the proposed framework.

References
Amelingmeyer, J. (2004), Wissensmanagement: Analyse und Gestaltung der Wissensbasis von
Unternehmen, 3rd ed. (in German), Dt. Univ.-Verl., Wiesbaden.
Argyris, C. and Schön, D.A. (1999), Die Lernende Organisation: Grundlagen, Methode, Praxis
(in German), Klett, Stuttgart.
Busch, P.A., Richards, D. and Dampney, C.N.G. (2001), ‘‘Visual mapping of articulable tacit
knowledge’’, in Eades, P. and Pattison, T. (Eds), Australasian Symposium on Information Visualization,
December 2001: Conferences in Research and Practice in Information Technology, Sydney, Australia,
3-4 December, Vol. 9, Australian Computer Society, Sydney, pp. 37-47.

Damerow, P. and Lefèvre, W. (1998), ‘‘Wissenssysteme im geschichtlichen Wandel’’, in Klix, F. and

Spada, H. (Eds), Enzyklopädie der Psychologie, Vol. 6: Kognition, Hogrefe, Göttingen, pp. 77-114.
Davenport, T.H. and Prusak, L. (1998), Working Knowledge: How Organizations Manage What They
Know, Harvard Business School Press, Cambridge, MA.
Davis, M.A., Curtis, M.B. and Tschetter, J.D. (2003), ‘‘Evaluating cognitive training outcomes: validity
and utility of structural knowledge assessment’’, Journal of Business and Psychology, Vol. 18 No. 2,
pp. 191-206.
Despres, C. and Chauvel, D. (2000), ‘‘A thematic analysis of the thinking in knowledge management’’,
in Despres, C. and Chauvel, D. (Eds), Knowledge Horizons: The Present and the Promise of Knowledge
Management, Butterworth-Heinemann, Boston, MA, pp. 55-86.
Dick, M. and Wehner, T. (2002), ‘‘Wissensmanagement zur Einführung: Bedeutung, Definition,
Konzepte’’, in Lüthy, W., Voit, E. and Wehner, T. (Eds), Wissensmanagement-Praxis, vdf Hochschulverl.
an der ETH, Zürich, pp. 7-27.
Drucker, P. (1993), Post-capitalist Society, New York, NY.
Eckert, A. (1998a), ‘‘Die ‘Netzwerk Elaborierungs Technik (NET)’ – Ein computerunterstütztes
wissensdiagnostisches Instrumentarium’’ (in German), Diagnostica, Vol. 44 No. 4, pp. 220-4.
Eckert, A. (1998b), Kognition und Wissensdiagnose: die Entwicklung und empirische Überprüfung des
computerunterstützten wissensdiagnostischen Instrumentariums Netzwerk-Elaborierungs-Technik
(NET) (in German), Pabst Science Publ., Lengerich.

j j
PAGE 32 JOURNAL OF KNOWLEDGE MANAGEMENT VOL. 11 NO. 1 2007
Eigler, G. (Ed.) (1990), Plato: Werke in acht Bänden. Bd. 6: Theaitetos, Der Sophist, Der Staatsmann,
griech./dt., bearb. von P. Staudacher (in German), Vol. 6, Wiss. Buchges, Darmstadt.

Engelkamp, J. (1991), Das menschliche Gedächtnis: das Erinnern von Sprache, Bildern und
Handlungen, 2nd ed. (in German), Verl. für Psychologie Hogrefe, Göttingen.

Engelkamp, J. (1998), Memory for Actions, Psychology Press, Hove.

Engelkamp, J. and Pechmann, T. (Eds) (1993), Mentale Repräsentation (in German), Huber, Bern.

Engelkamp, J. and Zimmer, H.D. (1994), Human Memory: A Multimodal Approach, Hogrefe and Huber,
Seattle, WA.

European Council (2000), ‘‘Lisbon European Council 23 and 24 March 2000: Presidency Conclusions’’,
available at: www.europarl.eu.int/summits/lis1_en.htm (accessed 24 September 2003).

European Foundation for the Improvement of Living and Working Conditions (2004), ‘‘Advancement of
the knowledge society’’, available at: www.eurofound.eu.int/publications/files/EF0409EN.pdf (accessed
4 June 2004).

FAW (2001), ‘‘Management von nicht-explizitem Wissen: noch mehr von der Natur lernen’’ (in German),
Forschungsinstitut für anwendungsorientierte Wissensverarbeitung, available at: www.faw.uni-ulm.de/
deutsch/publikationen/bmbf-studie/BMBF_Studie_Teil_1.pdf (accessed 20 May 2004).

Fitts, P.M. and Posner, M.I. (1979), Human Performance, Greenwood, Westport, CT.

Goldsmith, T.E. and Johnson, P.J. (1990), ‘‘A structural assessment of classroom learning’’,
in Schvaneveldt, R.W. (Ed.), Pathfinder Associative Networks: Studies in Knowledge Organization,
Ablex Publ. Corp., Norwood, NJ, pp. 241-54.

Goldsmith, T.E., Johnson, P.J. and Acton, W.H. (1991), ‘‘Assessing structural knowledge’’, Journal of
Educational Psychology, Vol. 83 No. 1, pp. 88-96.

Güldenberg, S. (1999), Wissensmanagement und Wissenscontrolling in lernenden Organisationen

(in German), Gabler/Dt. Univ. Verl., Wiesbaden.

Ichikawa, K. (1970), The Science of Creativity, Japan Broadcast Publishing, Tokyo.

Jonassen, D.H., Beissner, K. and Yacci, M. (1993), Structural Knowledge: Techniques for Representing,
Conveying, and Acquiring Structural Knowledge, Lawrence Erlbaum Associates, Hillsdale, NJ.

Klimecki, R. and Thomae, M. (2000), ‘‘Zur Bedeutung von Strukturmodellen für das
Wissensmanagement’’ (in German), paper presented at the Workshop der Kommission Organisation,
Zurich, 3-4 March.

Klix, F. (1984), Gedaechtnis, Wissen, Wissensnutzung (in German), VEB Deutscher Verlag der
Wissenschaften, Berlin.

Kluwe, R. (2004), ‘‘Informationsaufnahme und Informationsverarbeitung’’, in Zimolong, B. and Konradt, U.

(Eds), Enzyklopädie der Psychologie: Ingenieurpsychologie (in German), Vol. 2, Hogrefe, Göttingen.

Kogut, B. and Zander, U. (1992), ‘‘Knowledge of the firm, combinative capabilities and the replication of
technology’’, Organization Science, Vol. 3 No. 3, pp. 383-97.

Lee, S., Choi, K.S. and Choe, I.-S. (2004), ‘‘Representation of tacit knowledge in organizations’’,
available at: www.alba.edu.gr/OKLC2002/Proceedings/pdf_files/ID121.pdf (accessed 6 April 2004).

Li, M. and Gao, F. (2003), ‘‘Why Nonaka highlights tacit knowledge: a critical review’’, Journal of
Knowledge Management, Vol. 7 No. 4, pp. 6-14.

Maeda, A., Sugiyama, K. and Mase, K. (2002), ‘‘Media conversion of cyclic puzzles and development of
cyclic puzzle generators’’, IPSJ SIG Notes (Human Interface), Vol. 101, pp. 40-7.

Markowitsch, H.-J. (1992), Neuropsychologie des Gedächtnis (in German), Hogrefe, Göttingen.

Markowitsch, H.-J. (1999), Gedächtnisstörungen (in German), Kohlhammer, Stuttgart.

Markowitsch, H.-J. (2002), Dem Gedächtnis auf der Spur: vom Erinnern und Vergessen (in German),
Wiss. Buch-Ges, Darmstadt.

Nonaka, I. and Konno, N. (1998), ‘‘The concept of ‘ba’: building a foundation for knowledge creation’’,
California Management Review, Vol. 40 No. 3, pp. 40-54.

j j
VOL. 11 NO. 1 2007 JOURNAL OF KNOWLEDGE MANAGEMENT PAGE 33
 Nonaka, I. and Takeuchi, H. (1995), The Knowledge-creating Company: How Japanese Companies
Create the Dynamics of Innovation, Oxford University Press, New York, NY.

Opwis, K. and Lüer, G. (1996), ‘‘Modelle der Repräsentation von Wissen’’, in Albert, D. and Stapf, K.-H.
(Eds), Enzyklopädie der Psychologie (in German), Vol. 5: Gedächtnis, Hogrefe, Göttingen, pp. 337-432.

Polanyi, M. (1958), Personal Knowledge, Routledge & Kegan Paul, London.

Polanyi, M. (1966), The Tacit Dimension, Doubleday, Garden City, NY.

Polanyi, M. (1985), Implizites Wissen (in German), Suhrkamp, Frankfurt am Main.

Preece, P.F.W. (1976), ‘‘Mapping cognitive structure: a comparison of methods’’, Journal of Educational
Psychology, Vol. 68, pp. 1-8.

Quillian, M.R. (1968), ‘‘Semantic memory’’, in Minsky, M. (Ed.), Semantic Information Processing, MIT
Press, Cambridge, MA, pp. 227-70.

Rehäuser, J. and Krcmar, H. (1996), ‘‘Wissensmanagement im Unternehmen’’, in Schreyögg, G. and

Conrad, P. (Eds), Managementforschung 6: Wissensmanagement, de Gruyter, Berlin, pp. 1-40.

Renkl, A. (1996), ‘‘Träges Wissen: Wenn Erlerntes nicht genutzt wird’’, Psychologische Rundschau,
Vol. 47, pp. 78-92 (in German).

Rothe, H.-J. and Warning, J. (1991), ‘‘Zur Identifikation von semantischen Relationen im
fachspezifischen Gedächtnisbesitz von Experten vermittels Analogiebildung’’ (in German), Zeitschrift
für Psychologie, Supplement 11, pp. 375-84.

Scheele, B. and Groeben, N. (1984), Die Heidelberger Struktur-Lege-Technik (SLT) (in German), Beltz,
Weinheim.

Schindler, R. (2002), ‘‘Knowledge sharing by means of lessons learned repositories: how to assist
understanding, reflection and transfer’’, paper presented at the International Conference on the Future
of Knowledge Management, Berlin, 8-10 March.

Schnotz, W. (1994), Aufbau von Wissensstrukturen: Untersuchungen zur Kohärenzbildung beim

Wissenserwerb mit Texten (in German), Beltz, Weinheim.

Scholl, W. (2004), Innovation und Information: Wie in Unternehmen neues Wissen produziert wird (in
German), Hogrefe, Göttingen.

Schreyögg, G. and Geiger, D. (2003), ‘‘Kann die Wissensspirale Grundlage des Wissensmanagements
sein?’’, Diskussionsbeiträge des Instituts für Managements, Vol. 20, pp. 2-31 (in German).

Schütt, P. (2003), ‘‘Die dritte Generation des Wissensmanagements’’ (in German), available at: www.
km-a.net/file_upload/km-journal_1-03_schuett.pdf (accessed 11 April 2004).

Schvaneveldt, R.W. (Ed.) (1990), Pathfinder Associative Networks: Studies in Knowledge Organization,
Ablex Publ. Corp., Norwood, NJ.

Schvaneveldt, R.W., Durso, F.T. and Dearholt, D. (1985a), ‘‘Pathfinder: scaling with network structures’’,
CRL technical report no. MCCS-85-9, Computing Research Laboratories, New Mexico State University,
Las Cruces, NM.

Schvaneveldt, R.W., Durso, F.T., Goldsmith, T.E., Breen, T.J., Cooke, N.M. and Tucker, R.G. et al.
(1985b), ‘‘Measuring the structure of expertise’’, International Journal of Man-Machine Studies, Vol. 23,
pp. 699-728.

Shavelson, R.J. (1972), ‘‘Some aspects of the correspondence between content structure and cognitive
structure in physics instruction’’, Journal of Educational Psychology, Vol. 63, pp. 225-34.

Sinz, R. (1979), Neurobiologie und Gedächtnis: Neuronennetzwerke und Informationsspeicherung im

menschlichen Gehirn (in German), Fischer, Stuttgart.

Spender, J.C. (1996), ‘‘Making knowledge the basis of a dynamic theory of the firm’’, Strategic
Management Journal, Vol. 17, pp. 45-62.

Squire, L.R. and Frambach, M. (1990), ‘‘Cognitive skill learning in amnesia’’, Journal of Psychobiology,
Vol. 18, pp. 109-17.

Squire, L.R., Knowlton, B. and Musen, G. (1993), ‘‘The structure and organization of memory’’, Annual
Review of Psychology, Vol. 44 No. 1, pp. 453-95.

j j
PAGE 34 JOURNAL OF KNOWLEDGE MANAGEMENT VOL. 11 NO. 1 2007
Stehr, N. (1994), Knowledge Societies, Sage, London.

Strube, G. and Schlieder, C. (1998), ‘‘Wissensrepräsentation im Symbolverarbeitungsansatz’’, in Klix, F.

and Spada, H. (Eds), Enzyklopädie der Psychologie (in German), Vol. 6: Kognition, Hogrefe, Götingen.

Sugiyama, K. (2005), ‘‘Framework development and process analyses of organizational knowledge

creation in academic research projects’’, paper presented at the 1st World Congress of the International
Federation for Systems Research IFSR2005, Kobe, Japan, 14-17 November.

Sugiyama, K., Maeda, A. and Mizumoto, A. (2003), ‘‘Divergence via abstraction: practices for inventing
new puzzles’’, paper presented at the 4th International Conference on Systems Science and Systems
Engineering.

Sugiyama, K., Maeda, A., Osawa, R. and Mizumoto, A. (2005), ‘‘Creating new puzzles: practices of
‘abstraction and conversion’ strategy’’, Journal of the Japan Creativity Society, Vol. 8, pp. 1-20.

Tulving, E. (1972), ‘‘Episodic and semantic memory’’, in Tulving, E. and Donaldson, W. (Eds),
Organization of Memory, Academic Press, New York, NY.

Tulving, E. (1995), ‘‘Organization of memory: quo vadis?’’, in Gazzaniga, M. (Ed.), The Cognitive
Neurosciences, MIT Press, Cambridge, MA.

Tulving, E. (2002), ‘‘Episodic memory: from mind to brain’’, Annual Review of Psychology, Vol. 1 No. 53,
pp. 1-25.

Vogel, E.K., Woodman, G.F. and Luck, S.J. (2001), ‘‘Storage of features, conjunctions, and objects in
visual working memory’’, Journal of Experimental Psychology: Human Perception and Performance,
Vol. 27 No. 1, pp. 92-114.

von Glasersfeld, E. (1996), Radikaler Konstruktivismus: Ideen, Ergebnisse, Probleme, Suhrkamp,

Frankfurt am Main.

von Glasersfeld, E. (1997), Wege des Wissens: konstruktivistische Erkundungen durch unser Denken,
Carl-Auer-Systeme, Heidelberg.

Whitehead, A.N. (1929), The Aims of Education, Macmillan, New York, NY.

Corresponding author
Bertolt Meyer can be contacted at: [email protected]

To purchase reprints of this article please e-mail: [email protected]

Or visit our web site for further details: www.emeraldinsight.com/reprints

j j
VOL. 11 NO. 1 2007 JOURNAL OF KNOWLEDGE MANAGEMENT PAGE 35