Wandersee, J. H. (1990). Concept mapping and the cartography of cognition.

Journal of Research in
Science Teaching, 27(10), 923–936. doi:10.1002/tea.3660271002

Concept mapping is a metalearning strategy based on the Ausubel-Novak-Gowin

theory of meaningful learning (Ausubel, Novak & Hanesian, 1978; Gowin, 1981;

Novak, 1977; Novak & Gowin, 1984). It relates directly to such theoretical principles

as prior knowledge, subsumption, progressive differentiation, cognitive bridging, and

integrative reconciliation.

Basic to making a concept map for a piece of scientific knowledge is the ability

of the mapper to identify and relate its salient concepts to a general, superordinate

concept. That requires an understanding of what constitutes a science concept. Concepts

may be defined as regularities in objects or events designated by some label, usually

a term. Whether a process (e.g., precipitation), a procedure (e.g., titration), or a

product (e.g., carbohydrate), concepts are what we think with in science. Concepts

can be connected with linking words to form propositions (e.g., turtles are classified

as reptiles, sucrose tastes sweet, ontogeny recapitulates phylogeny). Therefore, a

concept map may be defined as “. . . a schematic device for representing a set of

concept meanings embedded in a framework of propositions” (Novak Lk Gowin. 1984,

p. 15).

At first glance, a concept map looks like a flow chart without the arrows. However,

it is not. Rather than representing a linear or branched sequence of steps in a process

or procedure, concept maps are designed to parallel human cognitive structure, in that

they show concepts organized hierarchically, whereas flowcharts do not. Instead of a

representation that corresponds directly to a linear text or lecture and reflects the

structure of knowledge (e.g., outlines), concept maps reflect the psvchologica/ structure

of knowledge. A concept mapper must often transform the knowledge to be mapped

from its current, linear form to a context-dependent hierarchical form. Before that can

be done, the mapper must first identify the key concepts, arrange them from general

to specific, and relate them to each other in a meaningful way

The advantages of mind mapping include its ‘‘free-form’’ and unconstrained structure. There are no
limits on the ideas and links that can be made, and there is no necessity to retain an ideal structure
or format. Mind mapping thus promotes creative thinking, and encourages ‘‘brainstorming’’. A
disadvantage of mind mapping is that the types of links being made are limited to simple
associations. Absence of clear links between ideas is a constraint. Mind maps have been said to be
idiosyncratic in terms of their design, often hard for others to read; representing only hierarchical
relationships (in radial form); inconsistent in terms of level of detail; and often too complex and
missing the ‘‘big picture’’ (Eppler 2006; Zeilik, nd). Mind mapping is also limited in dealing with more
complex relationships. For example, mind mapping might be useful to brainstorm the things that are
critical for students to recall in an exam (or a presentation, as in the example provided). However, it
is hard to see it being useful for a purpose that requires an understanding of how one concept is
essential to understanding another. More complex topics require more than an associational tool,
they require relational analysis. The tool of concept mapping has been developed to address these
limitations of mind mapping. Concept mapping Concept mapping is often confused with mind
mapping (Ahlberg 1993, 2004; Slotte and Lonka 1999). However, unlike mind mapping, concept
mapping is more structured, and less pictorial in nature. The aim of concept mapping is not to
generate spontaneous associative elements but to outline relationships between ideas. Thus,
concept mapping is a relational device. A concept map has a hierarchical ‘‘tree’’ structure with super-
ordinate and subordinate parts (primary, secondary and tertiary ideas). The map normally begins
with a word or concept or phrase which represents a focus question that requires an answer (Novak
and Can˜as 2006). Cross-links using connective terms (usually prepositional phrases) such as ‘‘leads
to’’, ‘‘results from’’, ‘‘is part of’’, etc., are used to show relationships between Fig. 1 A Mind Map
(‘‘Mind Maps Made With Mind Mapping Tool’’) High Educ 123 concepts represented. Examples (not
shown here) are added to terminal concepts as instances but these are not enclosed in boxes or
circles as they are not concepts but represent instances of a concept. Two quite different concept
maps are given below on the focus question: What is the purpose of concept mapping? Fig. 2. The
difference between mind mapping and concept mapping is also at the level of precision and
formality. Mind maps are less formal and structured. Concept maps are formal and generally more
tightly structured. Mind maps emphasise diagrams and pictures to aid recall of associations; concept
maps generally use hierarchical structure and relational phrases to aid understanding of
relationships. However, concept maps can take a variety of forms ranging from hierarchical, to non-
hierarchical forms, and even data-driven maps where the input determines the shape of the map.
One recent form of the latter involves a statistical process known as agglomerative cluster analysis
when analysis is made of terms that appear in a text across a number of respondents which are then
Fig. 2 Two different Novakian-style concept maps using the software CMap (http://cmap.ihmc.us/
conceptmap.html) (from ‘‘Concept Map,’’ 2010; Zeilik nd) High Educ 123 ‘‘clustered’’ to form a
diagrammatic representation (Jackson and Trochim 2002; Trochim 1989). A non-hierarchical, style of
concept map on the influence of labour market on the economy is given in Fig. 3. While non-
hierarchical, this map has more similarities to a concept map than a mind map as it endeavours to
establish appropriate relationships between the economic concepts rather than simple associations.
However, it has similarities to a mind map as well in terms of its looser, non-hierarchical,
unstructured form. The development of concept mapping has been attributed to the work of Novak
as early as 1972 and his work on children’s developing knowledge of science concepts (Novak and
Can˜as 2006). This work, in turn, was inspired by the work of learning psychologist Ausubel (Ausubel
1963). The mapping technique was refined further (Novak 1981) and then extended to the
educational context (Novak and Gowin 1984). The resulting diagrams are sometimes known as
‘‘Novakian maps’’ in honour of their founder. As noted, alternative approaches are also available
(Jackson and Trochim 2002). Recent additions to the Novakian format include attempts to capture
‘‘cyclical’’ relationships representing complex natural and social systems (Safayeni et al. 2005).
Technology has aided the popularity of concept mapping by means of dedicated software tools such
as CMap Tools (Can˜as et al. 2004) and Compendium.2 Such is the interest in concept mapping, an
annual international conference began in 2005. There are several stages in developing a Novakian
concept map. However, the stages are very different from developing a mind map: Fig. 3 Non-linear
concept map on labour market economics 2 Cmap Tools is available free from the Institute of Human
and Machine Cognition (http://www.ihmc.us). Compendium is available from the Open University
(http://www.labspace.open.ac.uk). A list of concept mapping software is available here (‘‘List of
Concept Mapping Software,’’ 2008). High Educ 123 1. Develop a declarative-type focus question (e.g.,
‘‘What is inflation?’’) 2. Devise a ‘‘parking lot’’ of concepts and ideas that are related to the concept
of inflation, and the question to be answered. The purpose of this stage is brainstorming. The
resulting concepts may or may not be used in the final map (Novak and Can˜as 2006). The concepts
are placed in circles or boxes to designate them as concepts. 3. Put concepts in hierarchical order of
importance in a provisional map. An ‘‘expert skeleton map’’ can be started by an instructor in a class
to scaffold the learning process, aid student participation and give students confidence. Students can
complete the map themselves with the focus question and concepts provided. 4. Link lines are then
provided between the hierarchical concepts from top to bottom. The conventions have changed over
the decades since the inception of concept mapping. Arrows were originally only used when it is
necessary to link a lower concept with a higher concept. However, this convention has recently been
revised by concept mappers to allow for arrows for all directions (Ahlberg 2004). 5. Devise suitable
cross-links for key concepts in the map. Verbs and prepositions/ prepositional phrases are used most
frequently, for example: ‘‘requires’’, ‘‘to work with’’, ‘‘will lead to’’, ‘‘involves’’, ‘‘during’’, ‘‘of’’,
‘‘through’’, and so on. The aim is to show the relationship between the key concepts and their
subordinate or super-ordinate elements. 6. Add examples to the terminal points of a map
representing the concepts. These are not enclosed in boxes or circles to delineate them as instances
of a concept. Since its inception as a formal technique, concept mapping has been widely used in
academic disciplines, for example, Accounting (Chei-Chang 2008; Irvine et al. 2005; Leauby and
Brazina 1998; Maas and Leauby 2005; Simon 2007; van der Laan and Dean 2006), Finance (Biktimirov
and Nilson 2003), Engineering (Walker and King 2002), Statistics (Schau and Mattern 1997), Reading
Comprehension (Mealy and Nist 1989), Biology (Kinchin 2000), Nursing (Baugh and Mellott 1998;
King and Shell 2002; Schuster 2000; Wilkes et al. 1999), Medicine (Hoffman et al. 2002; McGaghie et
al. 2000; West et al. 2000), Nursing (Beitz 1998) and Veterinary Science (Edmonson 1993). Research
has also been done on concept mapping as an assessment tool (Gouveia and Valadares 2004;
Jonassen et al. 1997; van der Laan and Dean 2006) and as a way to assist academics in course design
(Amundsen at al. 2008) and in managing qualitative data (Daley 2004). Several empirical studies have
ascertained the validity of the use of concept maps (Markham et al. 1994; Ruiz-Primo and Shavelson
1996). The main advantage of concept mapping is precisely its relational aim. Concept maps enable
relational links to be made between relevant concepts. In the educational context, it is claimed that
meaningful learning best takes place by linking new concepts to existing knowledge (Craik and
Lockhart 1972; Maas and Leauby 2005). Concept maps enable ‘the elements of [learning] to relate to
how cognitive knowledge is developed structurally by the learner’ (Maas and Leauby 2005, p. 77).
The main disadvantages of concept mapping are that they require some expertise to learn; they can
be idiosyncratic in terms of design; and because of their complexity they may not always assist
memorability, with learners faced with designing concepts maps often feeling overwhelmed and de-
motivated (Beitz 1998; Eppler 2006; Kinchin 2001). Others have noted that the rigid rules used for
identifying concepts and their multiple relationships does not make the process simple or easily to
learn, and the linear nature of concept maps mean that they are not adequate to capture more
complex relationships High Educ 123 between concepts. In particular, they do not enable easy
separation of concepts of critical importance from those of secondary importance (Daley 2004). It is
also impossible to distinguish identification of concepts from identification of arguments using a
concept map. For example, it is easy to imagine developing a concept map that canvasses the causes
and effects of the global financial crisis. In a complex issue such as this, multiple causes can be linked
to effects by means of relational arrows. A major disadvantage of concept mapping, however, is that
it is limited to relations between concepts. Many issues require more than an identification of
relationships between concepts; they require arguments to be made for positions that need to be
defended, and objections to those positions. For example, it is difficult to imagine how a concept
map could represent an argument for the claim that: ‘‘The US should have intervened earlier in the
global currency crisis’’. This kind of relationship is not, strictly speaking, relational. This is, of course,
not the fault of the concept mapping format. Concept mapping is a tool that was designed for a
different purpose. This is a limitation of concept mapping and it has led to the development of a new
kind of tool; a tool for mapping arguments. Argument mapping A relatively recent innovation,
developed since 2000, is computer-aided argument mapping (CAAM). Available in a wide-range of
software formats,3 argument mapping has a different purpose entirely from mind maps and concept
maps. Argument mapping is concerned with explicating the inferential structure of arguments.
Where images and topics are the main feature of associative connections in mind maps, and
concepts are the main relationships in concept maps, inferences between whole propositions are the
key feature of argument maps. ‘‘Arguments’’ are understood in the philosopher’s sense of
statements (‘‘premises’’) joined together to result in claims (‘‘conclusions’’). An example of an
argument map defending the proposition that The Reserve Bank will increase interest rates is given
in Fig. 4. At the first (top) level of the argument there is the contention. This is followed in this
example by a supporting claim (under the link word ‘‘because’’) and an objection (under the link
word ‘‘but’’). These are, in turn, supported by more claims of support or objection (which become
rebuttals when they are objections to objections): In the software, claims, objections and rebuttals
are coloured differently. Finally, basis boxes which provide defence for the terminal claims, are
provided at the end of the argument tree. Objections and rebuttals to objections can be added at any
point in the map (in different colours for easier visual identification). The ‘‘basis’’ boxes at the
terminal points of the argument also require evidence in place of the brackets provided. Some
evidence has been provided (‘‘statistics’’, ‘‘expert opinion’’, ‘‘quotation’’). Unlike mind mapping and
concept mapping, argument mapping is interested in the inferential basis for a claim being defended
and not the causal or other associative relationships between the main claim and other claims. The
software also allows for an automatically-generated description of the argument in text-form. In
some template argument formats—provided with the software—the mapping program also
constructs a prose version of the argument complete with a limited display of linking words.
However, this function is presently underdeveloped, and is a caricature of what would be needed in
university-style assignment. However, this impressive facility is indicative of where software tools are
headed. 3 Harrell provides a comprehensive list of argument mapping software (Harrell 2008). High
Educ 123 As noted, CAAM is still fairly new. Nonetheless, there have been several studies
demonstrating its impact on student learning, especially improvements in critical thinking (Twardy
2004; van Gelder 2001; van Gelder et al. 2004). Twardy demonstrated an improvement in critical
thinking skills as measured by a standard instrument in pre- and post-test by a 0.72 gain of standard
deviations. Van Gelder, Bissett and Cumming demonstrated an even higher gain of 0.8 standard
deviations in their study. A very recent study demonstrated greatest gains in students with the
poorest argument analysis skills in two separate studies over the course of one semester (Harrell
2011) The main advantage argument mapping may have over other forms of mapping tools is that it
focuses on a certain sub-class of relationships (i.e., logical inferences between propositions). It also
puts limitations around the items being mapped. There is a clear sense in which arguments—and not
relationships and associations—have ‘‘boundaries’’. Eventually, all reasons have to be grounded.
These grounds are presented as terminal ‘‘basis’’ boxes for assumptions. These are then evaluated
for plausibility as shown. With mind mapping and concept mapping, connections can potentially go
on ‘‘forever’’. A weakness of argument mapping is also its strength; argument mapping does not
capture looser, more tangential relationships, e.g., cause and effect. This makes it a tool with a very
precise purpose. However, as we shall see in the final section, there is no reason why the advantages
of argument maps cannot be supplemented with the advantages of other available tools, and with
additional refinements that do not exist at present. Another disadvantage of argument mapping is
that it can assume too much. In the educational context, argument mapping exercises can assume
that students have a sufficiently clear understanding of a topic or issue and the precise nature of the
task at hand. However, this understanding may often be absent. Students themselves may need to
define because because because but but because however The Reserve Bank (RB)will increase
interest rates. Inflation Inflation needs to be reduced. 2.9% too high The underlying inflation rate of
2.9% is too high. Web ABC news online Expert Opinion Macquarie Bank senior economist Brian
Redican CPI rising The Consumer Price Index (CPI) is rising at 1.9 %. Statistic Reserve bank of Australia
RBA website This rise is lowest in nearly eight years. Web ABC news online Election The RB will not
change interest rates during an election campaign. The Reserve Bank will be reluctant to influence
the outcome of the election. The RB Governor has said an election would not stop him. Common
Belief The claim is widely believed. Quote "If it's clear that something needs to be done, I don't know
what explanation we could offer the Australian public for not doing it, regardless of when an election
might be due." - Glenn Steven, Reserve Bank Governor ABC news Fig. 4 Argument map using the
software Rationale (http://www.austhink.com) High Educ 123 the scope of the issue to be addressed
and the exact parameters of the task. For example, faced with an essay topic as: • The changing roles
of men and women have been good for society. Discuss. Students may initially create a series of
arguments which implicitly focus on changes in their society, the society in which they are presently
living, or perhaps developed Western countries generally. They may never actually articulate what
the changes might be, or in what respects (or for whom) they might be considered ‘‘good’’ (nor
might they define what ‘‘good’’ means). They may not consider whether or not to confine
themselves to particular changes that have taken place over a particular time period in a particular
culture. Assignment topics are often deliberately ambiguous to allow students to demonstrate their
abilities in deconstructing the meaning of the topic itself. Working out what needs to do in an essay
and why is a preparatory, and a critically important step, to being able to map an argument
successfully. Students will have to do a considerable amount of initial reading and thinking and
struggle with key concepts before coming to an understanding of the exact task they need to
complete. It is only after this process that the student can map an argument. Argument mapping
software offers no help with these preparatory steps. However, this is precisely where a further
development in mapping technologies might be able to help (see ‘‘A convergence of mapping
tools?’’). Table 1 summarises the differences between the three forms of mapping discussed in this
paper. Notice that argument mapping shares the hierarchical form with concept mapping, and—in
some variants at least—argument mapping shares the design principles of colours, shading, and line
thicknesses with mind mapping. Note too the increasing level of sophistication in the tools. Where
mind maps have a high degree of generality in their application, concept maps are more specific
(focussing on relational factors) and argument mapping is the least general (more specific) in
application of all. This indicates, in one sense, some degree of perhaps unintended evolutionary
sophistication in the development of these tools. In the final section of this paper, suggestions will be
made on the new directions that this evolution might take. An important area of difference between
the mapping techniques is in the register and formality of language used, i.e., the differences in
linguistic ‘‘granularity’’ (see column to far right of table). Whereas in mind mapping the language is
fairly ‘‘loose’’, and can capture a variety of associative relationships, in argument mapping the
linguistic relationships are limited to whole propositions or statements linked by logical connectors
such as ‘‘because’’ or ‘‘however’’. Argument mapping requires precise rules of construction. This
forces explicit connections between propositions (from premises to conclusions or contentions).
Argument mapping thereby demonstrates a specific utility and considerable fitness to purpose. Mind
mapping does not have these constraints. Concept mapping occupies a space in-between the loose
and tightly constrained language in argument maps, and the looser, tangential, associative language
of mind maps. Concept maps typically involve the use of prepositional phrases such as ‘‘in relation
to’’, ‘‘is a result of’’, and so on; but, as we have seen, sometimes these rules are not adhered to.
Compare, for example, the very different examples of concept maps given earlier. The non-linear
economics concept map has elements of a more constrained mind map as well as having similarities
to a concept map