LINGUIST List 15.701

Thu Feb 26 2004

Review: Philo of Lang/Cog Sci: Shelley (2003)

Editor for this issue: Naomi Ogasawara <>

What follows is a review or discussion note contributed to our Book Discussion Forum. We expect discussions to be informal and interactive; and the author of the book discussed is cordially invited to join in. If you are interested in leading a book discussion, look for books announced on LINGUIST as "available for review." Then contact Sheila Dooley Collberg at


  1. Madalena Cruz-Ferreira, Multiple Analogies in Science and Philosophy

Message 1: Multiple Analogies in Science and Philosophy

Date: Wed, 25 Feb 2004 19:59:39 -0500 (EST)
From: Madalena Cruz-Ferreira <>
Subject: Multiple Analogies in Science and Philosophy

AUTHOR: Shelley, Cameron
TITLE: Multiple Analogies in Science and Philosophy
SERIES: Human Cognitive Processing 11
PUBLISHER: John Benjamins
YEAR: 2003

Announced at

Madalena Cruz-Ferreira, National University of Singapore


This book, derived from Shelley's doctoral dissertation, proposes a
first unified account of analogy by means of a formally constrained
model that treats single analogy as a special case of multiple

Chapter 1, ''The problem of multiple analogies'', gives a preliminary
layout of the theoretical framework from within which the analysis of
analogy will be approached, including purpose, methodology, and
reasons for the choice to focus on uses of analogy taken from biology,
archaeology and philosophy. The use of key terms is clarified, e.g.,
the terms ''analog'', which refers to both the source and the target
of an analogy, and ''multiple analogy'', an analogy in which ''_more
than one source analog_ is used to reason about a target analog''
(pp.3-4). Analogy, taken as a structured comparison between specific
items, is distinguished from induction, which involves generalisations
from observation of specific items. This importance of this
distinction lies in that taking multiple analogy as several single
analogies is erroneous, because it assumes that ''all analogies are
inductions'' (p.6), a view that has been found untenable in recent
research. Analogies are ''based on comparisons between causal
relationships in which the items in an analogy participate'' (p.6),
making causality, not universal generalisation, centrally relevant to
analogical reasoning.

The chapter also introduces discussion of modern theories of analogy,
which can be labelled ''shared-structure theories'' in that they focus
on the mappings, or ''alignment of corresponding parts'' (p.7), of
hierarchical causal relationships shared between source and target
analog. The rules and constraints of the mapping procedure are

Chapter 2, ''Multiple analogies and 'old fourlegs''', presents the
chosen analytical framework, the Multiconstraint theory of analogy
(MT), one version of shared-structure theory. Being a constrained
theory that allows application to both single and multiple analogies,
the MT appears as ideally suited to our understanding of the
undeniably vital contribution of analogy to scientific thinking. In
the MT, analogy involves alignment of conceptual structures ''in a
table of predicates of increasing abstractness'' (p.17). The
horizontal alignment of predicates with their arguments gives the
relationship between source and target analog, whereas the vertical
dimension accounts for a (causal) explanation that legitimises the

The chapter discusses the application of the MT first to single
analogies, and then its extension to multiple analogies. Examples are
taken mostly from evolutionary biology, particularly the claims
surrounding the Coelacanth 'living fossil', with a view to
systematising the rationale behind legitimate mappings between source
and target analog, as well as the criteria for analytical coherence.

The next three chapters, ''Multiple analogies from the Mesozoic'',
''Multiple analogies in archaeology'', ''Multiple analogies in Plato's
_Republic_'', discuss the uses of analogy in specific areas of
biology, archaeology and philosophy, respectively. These areas are
chosen precisely because analogy has played a major role in the
development of knowledge pertaining to each. A generalised account of
multiple analogies is progressively outlined, with rich discussion of
particular emblematic examples in each area, all illustrated with
tables detailing the horizontal and vertical steps and correspondences
that establish the analogy.

Chapter 6, ''Modelling multiple analogies'', presents Shelley's
proposed model, building on observations and conclusions gleaned from
the preceding, more expository chapters. Shelly discusses ''criteria
that a _general_ theory of analogy should satisfy'' (p.113), and
argues for a model which ''places much emphasis on purpose and
planning for the generation and understanding of multiple analogies''
(p.119). The planning involved in the construction of an analogy
naturally serves the purpose of that analogy. The model incorporates a
visual and a verbal buffer, accounting for the central role played by
visual/graphical imagery (''imagery augmented with a pencil and
paper'', p.50) and by language, particularly narrative modes (one way
of coding ''mental movies'', p.121), in the cognitive process of
analogical inference. The chapter's concluding remarks provide a
concise summary of the whole book.

An Appendix, ''Historical review'', gives a brief presentation of
treatments of analogy among Western thinkers, from Plato to
contemporary theorisation, with Aristotle's and J.S. Mill's accounts
described as the first examples of a model of analogy, and of an
explicitly constrained model of analogy, respectively. Shelley's
discussion groups analogy theories into three types,
shared-abstraction, shared-attribute and shared-structure, and focuses
on the theoretical treatment of multiple analogy.


Analogy is central to everyday reasoning about everyday happenings, as
well as to our attempts at interpreting and often predicting the
latter. Being a comfortable part of our household cognitive skills, we
hardly realise the extent to which it shapes our thought. It is by
analogy with last week's schedule that we plan next week's diary
entries, as it is by analogy with our human ways of being and thinking
that we conceive of life and intelligence beyond our species and our
planet. Analogy is a stepping- stone to approaching puzzles and making
them ours, by bringing them to a level that makes us understand
them. In this role of providing a bridge from a source analog (what is
known to us) to a target analog (what we want to know), analogy is a
product of our reluctance to live with unexplained phenomena, and
''clearly valuable in scientific thinking'' (p.12). Often there is no
other way to interpret evidence, particularly in research areas that
involve reconstruction of the past, whether in archaeology or
linguistics. Given that providing reliable explanations for puzzling
observations is the domain of science and of philosophy, and given the
thought processes that are common to both, the scene is set for the
analysis of knowledge garnered by means of analogy.

Shelley takes apart several well-established analogies and puts back
together, into one unified model, what makes these analogies
legitimate, or persuasive. By providing insight of this kind into the
workings of our mind, and into what exactly we select, along the
reasoning process, to make an analogy gain the sense we want it to
make, the topic of this book will appeal to a broad readership, not
only among biologists, archaeologists, philosophers and scientists in
general, but to anyone interested in the cognitive mechanisms of
acquiring knowledge.

Predictably, chapter 6 is the densest in the book, in that modelling
is a complex task both in the accuracy of the proposed representation
and in the fecundity of its workings. The task is of course not made
easier by the fact that a model is, arguably, itself an
analogy. Nevertheless, the text is overall clear and fluent, and
Shelley's claims are convincingly argued. The text is also clearly
signposted. All chapters begin with a preview of their contents, close
with a summary, and contain several tables spelling out the mechanisms
involved in particular analogies that are detailed in the text. The
style manages to combine precision with appeal, and the book at times
reads like an entertaining mystery novel. One example is the report on
the reasoning that led to the interpretation of the function of clay
figurine legs found in prehistoric Greek sites, given in chapter 4.

The book's appeal to the non-initiated suggests a comment about its
index, that could be put to better use in future reprints or
editions. As it stands, the purpose of the index, which includes
subjects and authors, is unclear. For some reason (modesty?) Shelley's
own name is absent from it, though quoted _passim_ in the text. The
page numbers given for several technical terms do not always list all
occurrences of these terms, nor their first or major occurrence -- if
'major' is indeed a criterion for index entry, its rationale should be
given. Other terms, that are central to the discussion and whose
indexing would therefore facilitate navigation of the text, include
'causality', 'predicate', 'argument', 'narrative', and 'false
analogy'. The latter, for example, refers to p.82, where its mention
in turn simply directs the reader to one of Shelley's own work's that
presumably provides definition and/or discussion of the term.

Shelley's model of analogy is formal, matching current trends in
cognitive science and related research areas that assume their objects
as formal objects, and therefore amenable to formalisation. Formal
models typically need primitives, from where inquiry into their
objects can proceed. But the model also encompasses the purpose and
the planning that presides over the use of analogies. Shelley's
discussion makes it clear that the purpose of analogies is that of
resolving the issue that made resource to them necessary in the first
place, whereas planning involves a series of measures whose collective
use satisfies that purpose. In other words, the primitives of the
analogy itself will vary according to purpose, be it explaining a
chain of events, or the function of a physical feature.

By highlighting that form cannot be independent from whoever is doing
the formalising, Shelley's model effectively (and sensibly)
incorporates the human factor in it. This point recurs along the book,
particularly concerning the ''different ways in which multiple
analogies may serve _cognitive goals_'' (p.32, emphasis added, MCF).
Analogies, like assumptions, are inevitably bound by the historical,
geographical, social and other factors that make up a scientific
paradigm, in Kuhn's (1970) sense of this term. They provide a way of
reading knowledge in order to gain more knowledge, both drawing on the
paradigm and in turn feeding it. This is so because analogies do not
just help provide plausible explanations, they must conform to a body
of acquired knowledge. For example, it is _plausible_ to conclude, on
the basis of anatomical analogy, that having two arms is consistent
with the hypothesis that human beings once could fly. But such a
hypothesis contradicts what we _know_ about flight, that appears to
have evolved from terrestrial gait, not preceded it.

Scientists, like all human beings, are nurtured by family, friends and
fellow scientists in their set of beliefs. It is this set of beliefs
that defines what can constitute a _finding_ that furthers our
knowledge within particular areas of research. As Shelley puts it, the
fecundity of a scientific practice is gauged by ''its ability to lead
to large numbers of _true beliefs_ for many practitioners'' (p.60,
emphasis added, MCF), and must therefore depend on nurturing. Since a
nurtured paradigm identifies what can be asked about a puzzle but,
more importantly, what cannot or need not be asked for a given
purpose, questions about the puzzle that might provide alternative
explanations are left unanswered and unanswerable, because they are
irrelevant for that particular purpose. For example, the accepted
construal of the split Greek figurine legs as two-part pledges, by
analogy with eg, the tearing in two of pieces of textile or paper as
proof of an economic transaction to be kept by each party (pp.71ff.),
satisfies the question about their function, but says nothing about
their form: why legs, and why women's legs? The analogical focus
directed at function that _explains away_ the figurine legs as split
tokens also means focus diverted from the form of the object, barring
the search for whatever significance might be attached to it.

Analogy ultimately depends on interpretation and argumentation, i.e.,
on what we humans think the observable can teach us about the
unobservable. Since cognition is necessarily an individual process,
and since individuals themselves obviously have preferences in their
ways of approaching knowledge, Shelley's model can be interpreted not
so much as a model of analogy as a model of analogical 'reasonings'
(with this word deliberately pluralized, MCF), that contemplates a
multidimensional view of cognition.

The quantification inherent in the workings of multiple analogies
point to a view of cognitive qualification, or categorisation, as an
additive property of analogy, where the only limit seems to be one of
''cognitive fatigue'' (p.54). Shelley argues that quantity does not
increase confidence in the analogy, nor therefore in the acceptance of
its authors' choice to use it. Legitimate source analogs are
independent from one another, and do their work on their own, due to a
source rejection specification such that ''the acceptance of one
analog would tend to suppress any consideration of its close
relatives'' (p.116). That is, similar sources cannot do the same type
of analogical work. On the other hand, a constraint on Shelley's
proposed model allows the mapping of predicate arguments according to
the similarity of their roles, not of their form. This is true of
analogical causality, in that predicates like ''in-order- that'' and
''because'' are equivalent: though the order of their arguments is
reversed, they both relate an antecedent to a consequent. An analogy
that correlates the two is deemed ''completely acceptable''
(p.17). That is, different predicates can be made to do the same type
of analogical work. This apparent contradiction is the result of
modifying one constraint of the model, by generalising its
formulation. This legitimises the analysis by making it flawless, a
condition to which formal thinking aspires. The issue is discussed by
Shelley, who concludes that problems such as these belong with the MT
itself, not with the specific analogies used by scientists.

The interplay of what is (dis)allowed in analogising, in terms of
legitimate equivalence, is what gives a measure of analogical
coherence, in which causality plays a central role. Coherence lies not
only in how well the analogy satisfies internal consistency, i.e., the
constraints of analogical reasoning, but is also ''a holistic property
of the context in which the analogy occurs'' (p.18), judged by its
good match to ''other things that are known or believed by the
analogizer'' (p.19). This is true of the good matches provided by
causality predicates. In addition, the role of analogical causality is
not so much that of establishing relationships among conceptual
constructs, and thereby organising thought, as that ''its mere
presence serves to increase likelihood'' (p.150). The cumulative
interplay among source analogs appears indeed to result in the
equating of the 'is like' of analogical comparison with the 'is' of
identity, in fact removing the distancing comparative connector: the
source 'is' a token of a particular type, and so 'is' the target.

Given that analogies themselves, and even metaphors, can serve as
source analogs (p.26), the one-time target analog is now in turn ready
to be used as a source, regardless of any fragility in its own
construal as a target. Is it in this sense that knowledge is
_transmitted_. Wright Mills (1967:405) noted the point, stating that
we ''live in second-hand worlds'' and arguing that ''experience itself
is selected by stereotyped meanings and shaped by ready-made
interpretations''. Transmission of knowledge proceeds by means of
intellectual ''memes'' (Dawkins' 1976 term), that are offered as
ready-made products, with no questions asked about their own
sources. Other individuals drink from the new source, and build
knowledge upon the results of the inferences that it may afford.

But knowledge must also be appropriated, if it is to play any useful
role. Like cognition, appropriation is an individual endeavour, and
the resulting knowledge can only be served (i.e., propagated) after
proper digestion (i.e., assimilation). The assimilation of knowledge
is facilitated in different ways, depending on the greater or lesser
appeal of the analogy itself. Vision, for example, is a very strong
cognitive mediator, and its central role in everyday cognitive
processes is duly incorporated as a visual buffer in Shelley's
proposed model. Visual substitution, in the form of analogical aids
like drawings, not only lies at the core of modelling reality, it
besides engages our ability to infer a process from still-image
representations. One example from the science of language that comes
to mind is the familiar theorisation about language features as
processes, or movement, aided by diagrammatic representations of
linguistic structure, that draw precisely on the primacy of visual
input among our senses.

The equally pivotal role of language itself in the process of making
sense of the world also finds its rightful place in Shelley's
modelling of a verbal buffer. Not simply because all human beings have
language, but because knowledge is useless if we cannot talk about
it. Again, the individual takes central stage, in that there are
persuasive users of language, and there are less skilled users.

Throughout the systematic analysis of different examples of analogy,
the book gives a fascinating account of the power of language in
shaping knowledge and, ultimately, in shaping thought. We are treated
to the _art_ of constructing knowledge, within the ''unended quest''
(Popper 1974/1992) that science itself is. We learn, for example, that
the reason why the familiar beast Triceratops has horns is that their
use was presumed through analogical correlations with the headgear of
modern hoofed mammals. This and other examples show that analogies
are ''flesh[ed] out'' (p.37) in various ways, in order to produce the
desired results. Ie, the reason is not _known_, but becomes
_knowledge_ by means of analogical argumentation. In addition, if
several things, not just one, are shown to be like something else by
means of multiple analogies, the familiar effect that transforms
quantitative information into qualitative knowledge takes place. The
features of each source analog need not be mapped onto the target
analog according to similar criteria of, say, form or function. The
analytical tables that Shelley provides for several examples of
multiple analogies show gaps in reasoning as blanks along the vertical
dimension of each table. These gaps are filled by implication (e.g.,
p.100ff.), that results from the cumulative effect of the horizontal
mappings of source analogs, regardless of the type and location of
their own gaps in reasoning. The parallel interplay of the different
sources forces the individual to fill these gaps, completing whatever
absent information they may stand for. Put another way, the gaps are
effectively disregarded, because their relevance becomes lost among
the sheer quantity of source analogs. Disregard is a cognitive process
that shares many similarities with cognitive normalisation processes
well-known since Bruner and Postman (1949), who showed that an inbuilt
property of human cognitive processes forces the perception of
incongruous stimuli as well-formed members of expected categories. As
in the barring of possible questions, discussed above, the act of
disregarding thus stands out as the crucial step in scientific thought
that it in fact is. It is in this sense that knowledge is
_constructed_ from information.

Shelley's purpose of modelling analogy is aptly fulfilled. The
discussion that supports the proposed model besides raises the
interesting question of the extent to which particular analogies, and
the knowledge that they serve, are themselves scientific, or a tool of
propaganda of ''scientific belief'' (Kuhn 1970:4), a phrase that is
oxymoronic only in appearance.

Being in more than one sense a manipulation, within what is allowed to
count as knowledge, analogies form a clever exercise in spin-doctoring
that favours the keepers of paradigmatic knowledge. It is then up to
the subscribers to alternative paradigms to deconstruct the analogy,
and expose the gaps that betray a disanalogy, an inference that lacks
coherent foundation. Disanalogies clearly ''serve the purposes of
one's professional critics'' (p.31). Given that analogising appears to
be a universal trait of human cognition, conflict must arise from the
professional (and social) goals to which particular analogies are
found more suitable. Choices among these surface in the lack of a
unified discourse, or a unified theory, that gives the appearance of
disagreement about knowledge, instead of disagreement about the way to
talk about knowledge. This is why the physical world, say, can be
explained by Newtonian, Einsteinian and quantum theories. This is also
why subscribers to different paradigms often perpetuate one another's
views as incompatible, in obedience to the ethical dictum ''help your
friends and hurt your enemies'' (p.30 & _passim_), instead of part and
parcel of the knowledge-gathering process.


Bruner, J. S. and L. Postman (1949). ''On the perception of
incongruity: a paradigm.'' Journal of Personality 18: 206- 223.

Dawkins, R. (1976). The selfish gene. Oxford, Oxford University Press.

Kuhn, T. S. (1970). The structure of scientific revolutions. 2nd
edition. Chicago, University of Chicago Press.

Popper, K. (1974/1992). Unended quest. An intellectual
autobiography. London, Routledge.

Wright Mills, C. (1967). The cultural apparatus. In I. L. Horowitz,
Ed. Power, politics and people: the collected essays of C. Wright
Mills. Oxford, Oxford University Press.


Madalena Cruz-Ferreira teaches linguistics at the National University
of Singapore. Her research interests include prosody, child
bilingualism and theory of science.
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue