Minds & Machines (2012) 22:263–269 DOI 10.1007/s11023-011-9267-6 BOOK REVIEW
Lorenzo Magnani: Abductive Cognition: The Epistemological and Eco-Cognitive Dimensions of Hypothetical Reasoning Springer, 2009, $157.00, ISBN 978-3-642-03630-9 Cameron Shelley
Published online: 27 December 2011 Ó Springer Science+Business Media B.V. 2011
The study of abduction has come a long way over the last few decades. A generation ago, the term abduction might evoke the following thoughts: 1. 2.
It is an argument in the form of the fallacy of affirming the consequent, and It concerns the logic, if any, of hypothesis generation in science.
Today, abduction evokes a broad and interdisciplinary study of a complex phenomenon, in which hypotheses play a central role. Magnani’s book is an ambitious and ecumenical essay on that phenomenon, an attempt to collect and integrate research, old and new, regarding abductive thinking. The author summarizes its aims succinctly: My aim is to combine philosophical, logical, cognitive, eco-cognitive, neurological, and computational issues, while also discussing some cases of reasoning in everyday settings, in expert inferences, and in science. The main thesis is that abduction is a basic kind of human cognition, not only helpful in delineating the first principles of a new theory of science, but also extremely useful in the unification of interdisciplinary perspectives, which would otherwise remain fragmented and dispersed, and thus devoid of the necessary philosophical analysis. (p. x) Professor Magnani is in an almost unique position to provide this analysis, as he has both contributed fundamentally to the literature on abduction, and also organized a compelling series of conferences on Model Based Reasoning over the last decade. As befits such a broad and involved topic, Magnani’s book is detailed and challenging. It is not an introduction: it does not attempt to ease the reader into the study of abduction, nor does it attempt to persuade a skeptical reader that abduction C. Shelley (&) Centre for Society, Technology, and Values, University of Waterloo, Waterloo, Canada e-mail:
[email protected]
123
264
C. Shelley
is a worthy topic of study. The audience of this volume is a readership already interested in the subject and acquainted with the basic issues surrounding it. Given the length and complexity of the text, the aim of this review is to outline the thread of its development for the benefit of a broader audience, and to help readers perceive the overall thrust of the work in the midst of the many details recounted in it. Chapter 1 characterizes and motivates a distinction, already introduced in Magnani (2001), between theoretical abduction and manipulative abduction. Roughly put, theoretical abduction concerns abductive cognitions that take place entirely within the mind, whereas manipulative abduction concerns abductive cognitions that involve external objects. In the former case, sentences or mental models are used as representations, whereas the latter involves the use and recruitment of writing, diagrams, and physical objects. In many respects, this chapter is an update and elaboration on the first chapter of Magnani (2001). Here, Magnani argues that both theoretical and manipulative activities can be understood as abduction if we consider them from a pragmatist perspective. In both theoretical and manipulative cases, the abducer is engaged in a project whose success involves arriving at a hypothesis. The only difference is whether or not the project is carried out entirely within the mind. From a pragmatist perspective, what matters is the success of the project, not how it is carried out. So, either theoretical or manipulative projects can be correctly considered as abductions. This view stands somewhat at odds with early descriptions of abduction in scientific discovery, in which only theoretical representations were considered. Thus, Magnani’s pragmatist perspective unifies the two sorts of activities under the rubric of abduction. Having unified different activities under the one concept, Magnani proceeds in chapter 2 to characterize what is essential to abduction. Breaking with what might be called received wisdom, Magnani disputes the notion that abduction is essentially about explanation. Instead, following Gabbay and Woods (2005), Magnani holds that the essential feature of abduction is that it is ignorance-preserving. The abducer begins, following Peirce, in a state of doubt or ignorance about some observation, e.g., why it is that monkeys love bananas. Recalling that animals love things that taste good to them, the abducer might hypothesize that bananas taste good to monkeys. Since this hypothesis could always be wrong, the abducer remains ignorant in the sense of having failed to achieve certainty. So, abduction is essentially ignorance-preserving. The abducer may proceed to submit the hypothesis to testing, which might show that it is better than any available alternatives (e.g., monkeys love yellow things?), which would be an inference to the best (available) explanation. However, the process of testing is inductive and so not essential to abduction. Of course, the abducer’s intention may be all along to submit the hypothesis to test, but that need not be the case. The remainder of the chapter insightfully applies this distinction between the generation and evaluation of hypotheses to a number of instances in science and mathematics. I am skeptical of the characterization of abduction as essentially ignorancepreserving. Ignorance is not a precise term but, in this instance, it seems to rely upon the isolation of knowledge involved in an abduction from all other knowledge of a cognitive agent. Consider the previous abduction again:
123
Book Review
265
Monkeys love bananas. All animals love things that taste good to them. So, bananas taste good to monkeys. Here it appears that the conclusion remains unevaluated in spite of its being a hypothesis resulting from an abduction. In reality, knowledge of monkeys, bananas and the like consist in a network of related concepts whose significance is determined, in part, from their position within that network. Abductions are constrained by the network, and this constraint constitutes an evaluation of the result. The conceptual links among monkeys, bananas, and taste suggest and constrain the abduction above, so that identification of the hypothesis is driven, in part, by its plausibility through the experience that configured the network in the first place. It is possible, of course, to create a formalism in which this fact is defined away, but the result is more a creature of the laboratory than a depiction of abduction in the wild. In any event, the contribution of Magnani’s book does not, in the end, rely on Gabbay and Wood’s approach, so the issue does not undermine its significance. In chapter 3, Magnani turns from the nature of abduction to its place in the development of the modern mind. He starts with the state of the prehistoric human brain, pre-adapted for symbolic cognition but still lacking the spark to touch off intelligent thought. Magnani posits that a feedback process developed between early humans and their material culture, that is, the external objects that they found useful for communication and survival. Through habits of mind developed in handling external aids like pictures and stone tools, early humans internalized physical manipulations of objects as internal manipulations of signs. He describes the process as the disembodiment of thinking, as it proceeds from concrete entities to virtual ones. Abduction enters into this narrative because it is, Magnani argues, particularly suited to being acquired in this process of internalization. Consider the use of diagrams to construct mathematical proofs: Diagrams are often drawn and then reinterpreted with the aim of identifying a visual or spatial relationship that could be exploited in a proof. The process of casting about for reinterpretations is abductive. Many other manipulations of external aids have the same reinterpretive character. As such manipulative abductions become internalized, then abduction itself becomes a normal mode of intelligent thought. Magnani finishes the chapter by arguing that external symbols and internal representations undergo a process of mutual refinement, whereby each is adapted to work efficiently with the other. Artifacts can act as ‘‘symbols that maximize abducibility’’ (p. 212), that is, creative reinterpretation. This mutual refinement is driven by the ecological niche that human beings occupy, that is, as consummate problem solvers. As abduction is key to problem solving, we alter our environment, and ourselves, to maximize our chances of abducing effectively. This chapter sets out the themes that Magnani explores in the remainder of the book. In chapter 4, Magnani explores the relation between abduction and action in the light of their deep connection in the natural history of the human mind. If abductive cognition is essentially manipulative in Magnani’s sense, then abduction and practical action should be intimately connected in the life of modern human beings.
123
266
C. Shelley
Reversing the approach of chapter 3, Magnani begins from the inside, with a consideration of abduction as a brain process. Abductive cognition is realized, Magnani claims, when one neurological process plays the role of hypothesis relative to another process. Watching a monkey relish a banana, for example, is realized as a multimodal pattern of neural activity, involving visual and kinaesthetic imagery, tracking the monkey’s appearance and eating motions. These activities, in turn, trigger other neural activities involving gustatory imagery, concepts, and emotions. The latter represent a kind of empathy, in which we imagine what the banana tastes like, the pleasure the monkey seems to derive from eating it, and how the pleasure affects the monkey’s outlook. Furthermore, the neural activities occasioned by some observation are not limited to tracking external events. They can also involve evaluations of the situation, and plans to intervene in it. Indeed, the connection between abduction and action goes very deep: ‘‘One of the central aims of abduction is to recommend a course of action,’’ (p. 234) and the adequacy of an abduction is often to be judged by whether or not it provides a hypothesis that is decisive when several alternative actions are possible. Thus, emotions and moral reasoning can be seen as intimately involved in abductive cognition because they provide the means through which evaluations and interventions are undertaken. Magnani draws some interesting conclusions from these considerations. For example, he calls for a re-evaluation of moral reasoning. On the received view, moral reasoning involves combining a set of moral concepts, e.g., consequences or duties, with the facts of a situation as they are known, from which a (or the) moral course of action is deductively derivable. Magnani argues for a view that begins with the facts as known and abductively identifies the moral concepts relevant to it. The best course of action, assuming there is one, is identified as part of this process. This view is different than the traditional one in Western philosophy in that it treats moral concepts naturalistically, and in that it allows for the moral justification of actions that cannot be deduced from those moral concepts. In chapter 5, Magnani expands the implications of his manipulative view of abduction by looking at animal cognition. His interest is not ethological so much as evolutionary, since animal cognition represents the foundation of human cognition in the distant past. So, Magnani explains the abductive nature of perception and instinct, both prime modes of animal cognition. That perception and instinct are abductive is not new; each is a kind of fallible and defeasible response to external conditions. There remains the issue of why perceptions and instinct often seem so good, that is, to capture the world more-or-less as it is. Peirce sometimes spoke of il lume naturale, referring to Galileo’s concept of the natural light in the human mind. Of course, this notion is both mysterious and human-centric and will not do. However, if perception and instinct are manipulative, as Magnani argues abductions fundamentally are, then they are constantly interacting with, and being constrained by, the external things that they relate to. Ontogenetically, this point implies that instincts and perceptions are not speculations but rather stages in a well-informed process of mutual interaction. Phylogenetically, it comports well with the fact that the perceptual and motor apparatus of animals represents not a blank slate but an interactive survival machine designed, as it were, in light of the experience of untold generations of ancestors, all of whom lived to revise and reproduce it successfully.
123
Book Review
267
Of course, as Magnani points out, to adopt this view, we must allow the concept of abduction to apply to processes other than symbolic thought or logical reasoning. Magnani argues his case by recasting not only perception and instinct but also conditioning and affordance as abductive activities as well. A critic might dispute Magnani’s abductive view by noting that perceptions, instincts, and so on, are often not defeasible in practice. For example, instinctive responses are involuntary for animals; they have not the wit to reject an instinctive response as inappropriate for a given situation. Magnani concedes the point but replies that it does not contravene his position: The instinct remains defeasible, that is, withdrawable in the face of suitable experience, even if the animal happens not to be physically capable of doing so. Similarly, you might say, the view out my window is paintable even if I happen to lack the skill to paint it. In chapter 6, Magnani solidifies this view by reconstructing the evolutionary process itself. Perhaps the default view of evolutionary history runs crudely along the lines of, ‘‘genetics proposes, nature disposes’’. In other words, random genetic variations within a species give rise to different phenotypes, on which natural selection then operates to determine which variations persist to be passed on. This view omits the possibility that animals may modify their environments to suit their genes, as it were. More specifically, Magnani claims, cognitive creatures tend to modify their environments to support their cognitions about it: ‘‘People alter and modify the environment by mimetically externalizing fleeting thoughts, private ideas, etc., into external supports’’ (p. 317). The environment that results from a long history of such ecological modification by people is a cognitive niche. The mutual accommodation that brings about a cognitive niche occurs manipulatively through adjustment of, and adjustment to, the affordances of that environment. Magnani argues that this process applies especially and distinctively to human cognition. We are the species that, more than any others, has created a cognitive niche in our ecology. This line of argument would seem naturally to lead to some conclusions about the role of technology in human development, but Magnani settles on the development of abstract models of the world such as geometry. Of course, technology in this role has already been examined in chapter 3. It is interesting to consider how Magnani’s ecological characterization of abduction might change old disputes about the well-foundedness of human knowledge of the world. In his Meditations, Descartes doubted the veracity of his perceptions of his room and even of his notions of geometrical figures. Only a benevolent God could guarantee their correctness, he concluded. I can imagine a response from the perspective of the human cognitive niche: Descartes’ perceptions of his room are founded, in no small part, on a long history of mutual adaptation between artifactual design and perceptual apparatus. A similar, albeit more complicated story could be given about his notions of geometry. No need for a benevolent God as much as a long history of designing ancestors. Magnani spends chapter 7 exploring some of the implications of his ecological view of abduction in human cognition. There is, for example, an old and ongoing discussion about the relation between formal logic (and mathematics, and computation) and human psychology. ‘‘Does formal logic constitute a set of laws of thought?’’, to put the matter crudely. Magnani argues that logic results from
123
268
C. Shelley
externalization of thinking, which is then re-internalized. In brief, people have constructed external models of their own thinking patterns and have abductively refined them into different logics suited for different purposes. These refinements are then internalized as ‘‘idealized logical agents’’ (p. 384), that is, good thinkers who people may emulate to become good thinkers themselves. Magnani explores some of the limitations of idealized logical agents by examining the role of logical fallacies in practical reasoning. Hasty generalization, for example, is a fallacy from the perspective of deductive logic because it is not truth-preserving. However, such fallacies may make sense from an ‘‘eco-logical’’ perspective, especially when they facilitate the preservation of good social order. Magnani notes, for example, that fallacies such as ad hominem, ad populum, straw man, etc., play a significant role in the formation and communication of gossip (p. 410) and other forms of social assessment. For example, some time ago I listened to a researcher in a think-tank discuss his view of a proposed government policy (he was critical of it). The radio interviewer mentioned that the think-tank received funds from political opponents of the government. The researcher testily responded, ‘‘I fail to see the point of that ad hominem.’’ He saw his affiliation as irrelevant, a fallacy serving only to distract attention from the merits of his argument. Of course, the fallacy invites the audience to consider that he and his argument may be biased. As Magnani points out, it is hard to see how human beings could function in a complex social world without making (judicious?) use of patterns of reasoning like the ad hominem. So, idealized logical agents in which ad hominem is poor thinking are limited in their scope of application. Idealized logical agents in which ad hominem may be considered sound (abduction) should not be regarded as poor cousins of the first. The final chapter provides an exploration of two further issues: (1) the dynamics of abductive cognition in groups and (2) the role of abduction in the formation and regulation of the mores of social groups. On the first point, Magnani presents a vocabulary for describing how abductions arise. In brief (and if I have it correctly), abductive hypotheses emerge from a process of ‘‘gluing’’ together adumbrations or partial hypotheses achieved by examining a problem from different perspectives. The language proposed is an amalgamation of dynamical systems theory (‘‘attractors’’, ‘‘state space’’, etc.) and phenomenology (‘‘adumbrations’’, ‘‘pregnance’’). Magnani gives, as an example, the discovery of Uranus in which the hypothesis that Uranus exists (from Adams and Leverrier) served as an attractor that motivated Galle to point his telescope upwards to observe the planet, thus causing a reconfiguration of the understanding of the solar system (p. 432). Although Magnani does not mention Kuhn, the result sounds much like an account of Kuhnian paradigm shifts. Magnani notes that this account of abduction remains metaphorical, so we must await future work to see new things it will reveal about abduction. In the final part of the chapter, Magnani continues the discussion from previous chapters about abduction and the human cognitive niche. He contends that the human mind (and its environment) has been profoundly shaped by the need of people to work together by forming coalitions. Coalitions facilitate the sharing of information needed for the abductive niche but must sometimes be maintained by violence. This fact is reflected in human nature itself:
123
Book Review
269
In this perspective, the long-lived and abstract human sense of guilt represents a psychological adaptation, ‘‘abductively’’ anticipating an appraisal of a moral situation to avoid becoming a target of coalitional violence. (p. 448) So, guilt is a kind of internalization of a standing threat from your social group to punish you for not toeing the line. This view of human nature is quite thoughtprovoking: Before fire or clubs or sticks, human beings formed and used each other as a kind of technology for grasping and getting along in the world, with the result that we have fundamentally designed ourselves as cognitive and moral agents. The take-home message from Magnani’s book is quite a welcome one: Research in abduction is not the study of a logical fallacy but is, instead, an inquiry into human nature. I look forward to the next provocative and well-researched installment from Magnani’s prolific keyboard.
References Gabbay, D. M., & Woods, J. (2005). The reach of abduction: A practical logic of cognitive systems. Amsterdam: North-Holland. Magnani, L. (2001). Abduction, reason, and science. Processes of discovery and explanation. New York: Kluwer.
123