2.5.2  Types of reasoning 

A  Pragmatic View of Reasoning

We have the ability to reason about ourselves and our environment and do so incessantly every day of our lives in order to act in the ways that sustain our existence on this planet. When we begin to ask questions about the nature of reason itself (i.e. when we reason about reason) we have entered the domains of logic, cognitive psychology, and philosophy of mind.

From the perspective of neo-pragmatist philosophy, the aim of good reasoning is to arrive at useful descriptions, explanations and predictions that enrich our worldview and enable our goal-mediated actions. In biological terms, good reasoning facilitates the complex adaptive behaviours of animals such as ourselves and so makes our survival as a species possible in a wide range of ecological niches.

The conclusions resulting from our introspection, practical investigations and intellectual enquiries do not need to lead to unalterable and unquestionable truths in order to be termed good or useful. Nevertheless, it is legitimate that, at times, we enjoy the feeling of certainty, and we benefit from the practical rewards that emerge from sound reasoning.

If we are to produce good analysis of empirical observations and to develop predictively valid explanations we need to rely on our ability to reason in a worthwhile way. The basis for our ability to reason rests on our biologically innate capabilities, our cultural indoctrination and the capacity that we might have for imaginative or original thought. Good styles of reasoning come in many forms. Whether or not they all deserve the descriptive term ‘logical’ is a moot point, since we do not routinely seek to legitimise our arguments via a limited set of axioms that underpin formal logic ( see the section on axioms).

An Introspective Characterisation of Types of Reasoning

Many styles or types of reasoning processes have been identified.  Although some types (or styles) of reasoning are listed below for the sake of simplicity and convenience, it is worth remembering that they do not operate independently of one another. Indeed it is likely that this classification is merely an abstract, introspective and reductionist approach that helps us identify characteristics of very complex processing or reasoning that includes the influence of previous experience and current sensory input, where that is appropriate. The abstract characterisation below is not intended to be comprehensive but instead to indicate some of the rich and varied ways in which we all reason, without making any reference to the underlying biological processes. 

• Practical, Procedural and Embodied Spatial Reasoning  Practical reasoning underpins the decision making that facilitates much of our physical actions. Practical reasoning also encompases diagnostic reasoning (see the video below) that is required in fields as disparate as medicine and electromechanical engineering. Procedural reasoning can be both practical and conceptual and examines the ‘how’ (involving causes) and the ‘why’ (involving reasons) of taking action. It is usually envisaged as a series of steps from physical cause to physical effect or from psychological reason  to justified action. Embodied spatial reasoning is our cognitive ability to sense and manipulate our own bodies and interact with the 3-dimensional world  around us. These are by far the most important styles of reasoning for the pragmatist who seeks to operate outside the constraints of classical epistemology, where ‘definitively knowing’ has been taken to almost obsessional levels.  Embodied spatial reasoning for example is certainly about what ‘works’ (and does not work), although it might be described in declarative terms.
• Associative Reasoning  Psychologists have developed the idea of associative memory, thinking, or learning in addition to traditional philosophical views of rationality. Associative thinking is sometimes considered to arise from perception and thus experience. If taken seriously, we can see that associative thinking could form the basis of many human cognitive abilities. Why am I so sure, for example, that hippopotamuses cannot fly? There are lots of characteristics that I ‘associate’ with flying creatures, such as the possession of wings, which are absent from the hippopotamus. There are also characteristics of these animals that definitely are not associated with flying such as having 4 legs and a mass too large for biological winged flying or gliding.
•  Relational Reasoning  A commonly recognised and widely used style of reasoned argument is inference by analogy (also referred to as Analogical Reasoning).  This is an example of a type of relational reasoning. Analogy is used in legal practice to make arguments, for example, when citing case law as being relevant to the circumstances currently under consideration. A historical example in physics was the errant analogy made between the energy levels of electrons bound to nuclei when they were compared with planets orbiting our solar system. Although there might be a close linkage between association and analogy, the latter can, at times, be considered to operate in more contemplative and abstract domains of thinking. The same is the case for transitive relational reasoning. Transitive reasoning is often exemplified as: If A is taller than B and B is taller then C. Then A is taller than C. Transitive reasoning is the ability to form useful inferences based on indirect relational evidence. 
• Social and Ethical Reasoning  As social animals it is essential that we reason about how to treat our biological kin and the wider communities in which we live. In complex modern societies we need these forms of reasoning to be codified.
• Numeric Reasoning One of the most fundamental developments in human culture was the invention of numbers. Reasoning based on numbers is widespread in many fields of research and practice including mathematics, statistics and probability theory, the natural sciences, engineering, medicine and the applied sciences, the social sciences, economics and finance, and the law.  
•  Formal Spatial Reasoning Geometry and topology are formal expressions of our ability to reason spatially and are based on sets of axioms.
• Induction is the imaginative creation of generalised descriptions and explanations inferred from single or repeated observations. The degree of generality of an inductive assertion can vary between being locally useful and some attempt at a much more widely applicable law-like explanation. Induction, when used wisely, can generate strong or good explanations that robustly hold over the course of extended inquiry. The prime outputs of induction are description, explanation, prediction or practical utility within a context, not truth in an idealised epistemic sense. The conclusions of induction are often viewed as probabilistic rather than certain. Of course, this begins to sound like deductive arguments in which we have substituted truth for possible truth and so have produced an output that is less than certain. Inductive arguments are, by their definition, not deductively valid. Seeing that as a problem is, of course, nonsensical.  Despite many arguments to the contrary, for the pragmatist there is no ‘problem of induction‘, since it is by combining different forms of argument or styles of thinking that we learn and thus arrive at acceptable and predicatively useful explanations of the world.
• Mechanistic Reasoning  Our understanding of complex entities is made simpler by understanding the parts, properties and interactions of components of complex systems. Understanding how the  components of a system interact is fundamental in astronomy, chemistry, biology and medicine, for example. It is common for us to refer to mechanisms of disease or the mechanisms of drug action when we seek to understand the molecular nature of pathological processes and treatments.  When we interpret the passage of the sun across the sky each day as the spinning of the earth around its axis of rotation we are reasoning mechanistically rather than thinking of the repeated occurrence as a (near) certain induction. Mechanistic thinking is an imaginative process that underpins much of science both pure and applied.
• Abduction is the attempt at construction of the best or most likely explanation available given certain observations and beliefs. In philosophical jargon this way of reasoning is known as Inference to the Best Explanation.  This form of inference is widely held to be common in both scientific practice and in everyday life.  Inference to the Best Explanation is local in nature rather than general in scope. Whether or not this type of reasoning is ‘logical’ in any formal sense is debatable, for it is difficult to conceive of any set of  axioms that would act as the basis of such a formality. As far as I can see no set of axioms is even possible since the term ‘best’ is ill-defined, context specific, and restricted by our cognitive capacities. The personal satisfaction obtained in imagining our best explanations should not, of course, be confused with deciding that we have reached an understanding that might be close to an idealised truth.
•  Predictive Reasoning After we have formulated ideas by the styles or reasoning mentioned above the pragmatist will want to test their empirical value by making mechanistic predictions. Successful predictions would then be seen as supportive of a particular set of ideas. Unsuccessful predictions would then highlight a need for re-evaluation.
• Logical  Deduction is a style of argument or reasoning in which conclusions are reached given 2 or more initial assertions (also called premises). Deductive Logic is concerned with the form or structure  of arguments. Logic, in a strict sense, is about the validity and soundness rather than the truth of reasoning processes since deduction has nothing to say about the input truth value of the initial premises of an argument. Deduction is at best said to be ‘truth-preserving’ when the way in which the premises are connected in an argument is ‘validly’ structured. We should take this property of deduction very seriously since valid deduction preserves rather than generates truth (or a pragmatic alternative such as acceptability).  If we make the supposition that the premises are indeed true, the outcome of a validly structured argument is said to be ‘sound’. In other words if an argument has true premises and is validly structured the resulting argument is said to be ‘sound’. However by convention, if we assume the truth of the premises of particular instances of logical deductive argument then we arrive at the binary outcomes of truth or falsity. This is because ‘logical truth’ or falsity convey the notion of whether or not in a particular instance an argument is accepted or denied. More precisely speaking, we are using the notion of logical entailment (that will be discussed later)

Clearly, there may be domains of explanatory inference where only one type of argumentative process applies. The formation of new ideas in some branches of logic, for example, might be the closest we ever come to a purely deductive process.  In the case of logical deduction, the premises might  have arisen at least in part by induction and the conclusion through the deductive process. Clearly a particular Inference to the Best Explanation could involve associative reasoning, analogy and induction etc.

In matters of formal logic we can substitute the binary outcomes of truth and falsity with some other formalism for believing or accepting the conclusion and still claim to be asserting a form of rational thinking. If however in the long-term, you find that you have no reason for believing an apparently sound deductive conclusion, you might then wish to question the basis for believing the premises. For the pragmatist it is essential that we are able to make revisions of beliefs when it becomes apparent that our thinking lacks coherence. In technical terms, this way of reasoning is encapsulated in non-monotonic logic.

Complementary Views

In addition to using an abstract, introspective, reductionist, and psychologised approach to reasoning, in which we sketch out aspects of thinking through introspection and the analysis of language use, we can also take to a more biological view involving multi-modal sensation, perception, learning and memory.  We might be even more reductionist and examine neuronal communication and develop models at that level. If, for a moment, we ignore the present physical impossibility of probing the human brain at that level, it seems fundamentally unlikely that we could recover the sort of introspective classification set out above. The easiest way to explain that idea is perhaps to use analogy. Understanding brain function at the level of the neurones might be somewhat equivalent to trying to understand the output of a very complex computer system at the level of individual logic gates that carry out the Boolean operations on the binary logic of 0’s and 1’s. The high degree of emergence in biologically complex systems (explained in another section) makes description at various levels of abstraction perfectly legitimate, especially for the pragmatist.

We could express an understanding of brain function at a more abstract level in terms of control theory. In this way the idea of feed-back and feed-forward neural control loops could be invoked. “Feedforward loops predict what is going to happen, while feedback loops confront the prediction with what happened so that we can react accordingly.” [See source >]  It is believed that fine control of bodily movements by the cerebellum operates using such loops. In addition the presence of this control system might contribute to our feelings of agency.

Such control systems are thought to be very advantageous for immediate and rapid control of limb movement. However, as yet, it is unclear why they would be beneficial for abstract forms of cognition in which there is no evidence of direct sensory regulation being involved in ‘error’ detection. As the history of philosophy and science demonstrates, the development and critique of ideas can be an exceptionally slow process in which changes in language, mathematics, theory and technology and wider culture play their part. Consider the extreme example, of how ideas about the existence and structure of atoms started in ancient Greek times and then were slowly re-developed during the ‘Chemical Revolution‘ and up until the present day. In such abstract matters we do not seem to operate the classic feedback and feedforward control systems. Instead we have to use a painstakingly slow form of reasoning, which in my case often takes years to complete or even modify. The dichotomy between fast and slow relational inference has led one group of psychologists to claim there is a distinction between perceptually driven fast thinking and a slower reasoning process.

One very interesting level of abstraction used in constructing a narrative about how we reason is to be found in computational theories of mind ( also see the IEP article on CTM). If we view the central nervous system and its sensory apparatus as a biological implementation of a conceptual Turing Machine, the idea of computation is coherent with our wider understanding of the physical world. For the neo-pragmatist, epistemic coherence is virtuous because it can, at least in part, act as the basis of justification of ideas.

The idea of a computational theory of mind has been inspired by the concept of ‘machine learning’ and more recently by the advent of so-called generative artificial intelligence in computational models of language usage, visual imagery, and robotic control systems. In these ways we can develop narratives, and conjecture an information theory-based view of how we deal with the world, based on the metaphysical idea of biologically encoded information. The British cognitive philosopher Andy Clark, for example, argues that the embodied brain creates a predictive model of ourselves and the external world that we constantly update though ‘error’ detection mechanisms. (Watch the embedded video below for a useful synopsis by Clark himself.) These newer ways of thinking will, in time, generate a much richer and much more complex understanding about the ways in which we reason.

Of course, the idea that prediction is central to our understanding of the world is already part of the more abstract pragmatic justification of ideas that we see in science, epistemology and in ordinary living.

Version 4.3

---

 

Relevant Videos

A Video Illustrating Diagnostic Reasoning in Electromechanical Engineering

This video makes it very clear how a great deal of learning and deduction can be required to master the art of diagnostic reasoning. In this video we can watch a true master at work, within his domain of expertise. A really helpful feature of this video is the very clear description of the logic used to motivate the actions of the person. Any analogical reasoning in this situation must be obtained from previous experience and learning. A procedural approach is explicitly needed to identify the cause of several malfunctions. Early parts of the video also demonstrate exquisite spatial reasoning.

Videos About the Embodied Brain as a Predictive Organ

 Andy Clark advocates that prediction, error detection, and learning are important in reasoning. He also takes the interesting view that this way of thinking about the way the embodied brain can also provide insights into consciousness (See the video below)

< previous | index | next >

Copyright 2024, Steve Campbell