The capacity for human reasoning is extraordinary. Reasoning involves making suppositions and inferring their consequences. But what makes great thinking so difficult is having the imagination to ask the right questions, or the ability to see things from a completely new perspective.


Philosophers distinguish between 3 main forms of inference: deduction, induction, and abduction. Abduction means “reasoning to best explanation” and medical diagnosis is a good example of this. Such inferences are generally probabilistic — the best explanation is the most probable, but not necessarily correct.


Deductive or logical inferences are assigned an especially important status by philosophers. The reason is that a logically valid argument will guarantee a true conclusion given true assumptions. Most mathematics can be regarded as a special form of logic, in which theorems are deducted from axioms or assumptions.


In logical systems there are always some things postulated or assumed from which everything else is deduced. These axioms are typically intuitive and cannot themselves be proved. If there is a limitation, it is here that it will be found.


Even if someone’s reasoning is sound, we may not agree with their conclusions because we do not accept their premises. That is why political arguments may appear futile or irresolvable. Typically, such debates take place between people with different belief systems. No amount of logical reasoning will resolve their argument. Logic works best in scientific domains where people can agree on some basic assumptions and practices.


If we only had deductive inference to call on, then we could never learn anything new. Deductions merely bring out conclusions that follows from what we already believe or assume to be true.


Philosophers from Aristotle onwards were interested in trying to understand and explain human thinking. It is, after all, the main tool of their trade. The ancient also believed, as most psychologists still do today, that it is thinking — or more accurately a certain kind of thinking — that separates human intelligence from that of all other animal species. Animals generally learn to adapt their behavior on the basis of what has been successful in the past. Only humans can imagine the future, or alternative futures, and then calculate the course of action needed to ensure the best outcome.


This book is called Thinking and Reasoning because most of the kinds of thinking studied by contemporary psychologists involve thinking that computes some kind of outcome and can be described as one or another kind of reasoning.


James was worried that people could only report their thoughts as memories of something past and that introspection was itself a mental act. Asking someone to report their thoughts would surely change what they were thinking.


His more profound observation was that the production of these ideas was automatic, operating beneath the level of conscious thought: “The more I have examined the working of my own mind, the less respect I feel for the part played by consciousness. Its position appears to be that of a helpless spectator of but a minute fraction of a huge amount of automatic brain work.” We do not have access to how our minds work simply by studying our own consciousness.


Modern psychology is dominated by the scientific method, in which theories must be tested empirically, preferably by controlled experiments. A scientific theory must, in principle, be open to falsification. Psychologists do not regarded Freudian theory as scientific in this sense.


If thinking is the flow of consciousness, then the mind is conscious. But Freud and other psychoanalysts argued that our behavior is often controlled by an unconscious or preconscious mind. Freud believed that consciousness was the tip of the iceberg. His primary interest was in the treatment of mental disorders such as free-floating anxiety, phobias, and obsessive-compulsive behaviors. The Freudian unconscious is the repository of repressed emotions which can lead to irrational behaviors. Associated with repression are defense mechanism, like projection and rationalization.


Rationalizations: false reasons invented by the conscious mind to explain behavior caused by the unconscious.


The fact that people can happily rationalize explanations for their own behavior is another black mark for the introspective method.


The tricky side of this is that people don’t know what they don’t know. One of the basic rules of psychology is that people will answer any damn silly question you put to them. That does not mean you should believe the answers.


Early in the 20th century, a movement called behaviorism was founded in strong reaction against introspective psychology. Behaviorism banished any reference to the internal workings of the mind and was to dominate psychology for the best part of 50 years. The rationale was clear. Psychology is a science and as such must rely on objective observations which can be repeated by another observer. On this basis, we can only study behavior and not mental states.


The study of the human mind has re-entered psychology in a different form than that which preceded behaviorism. Thinking is no longer seen as a flow of consciousness, but rather as a high-level form of information processing in the brain.


Ill-defined problems may be quite easy for a human to solve but would be next to impossible for a computer, unless it knew all the things that we know. By contrast, some problems are well-defined. This means that there is a clear set of rules that can be applied to get from where you are to where you want to be. Artificial problems usually have this nature. If a problem is well-defined, a computer program can in principle be written to solve it.


They tried to explain all human intelligence in terms of conditioning and habit learning, as though human behavior could be explained simply by studying the learning of rats and pigeons. On this basis, all human problem solving would have to occur gradually by trial-and-error learning.


The term gestalt is used to refer to the idea of “good form” or whole figure, whose properties cannot be deduced as the sum of its parts.


One major issue in this period focused on whether problems were solved in a gradual or incremental manner (as claimed by behaviorists) or by a sudden and discontinuous insight (as argued by the Gestalt psychologists). Insight problems are those solved with a sudden “Aha!” experience, although not all problems work this way of course. However, the discovery of insight problems show that problem solving involves more than habit learning.


Both humans and computers needed intelligent methods to reduce the search space, and called these heuristics. Heuristics cannot guarantee success but they can lead to effective solution to problems that are otherwise computationally intractable. One method is to work backwards from the desired solution to the current state, creating a subgoal. Another is means-ends analysis, where one calculates the difference between the current state and the goal state and tries to reduce that difference.


People do indeed use heuristics to help solve problems with large problem spaces. There are many heuristics that humans can use when faced with difficult problems but they will not necessarily be applied consciously. Heuristics can operate in an intuitive or perceptual manner, as when strong chess players just “see” better moves than weaker players. This comes from a great deal of experience, however, and is a kind of insight, as potential solutions are perceived without any process of conscious reasoning. Heuristics can also be explicit rules that can be taught to people to aid their problem solving.


You might think that this means experts are more intelligent, but this is to misunderstand the nature of expertise. What marks experts out is the extent of specialized knowledge that they have acquired. This knowledge is often manifest in the form of superior intuition or Type 1 processes. Experts rely on pattern recognition. From experience, they recognize situations that are similar to those seen before and can rapidly propose potential solutions that have worked in the past. Only when problems are unusually novel that they need to engage in explicit reasoning of a Type 2 nature.


Research on problem solving shows that there is more to it than reasoning or computing solutions to well-defined problems. Some problems are hard because there are too many possibilities to consider and some shortcuts must be found to solve them. Others are hard because we simply lack the right approach and mental set and may only solve them when a rest, a hint, or an analogy enables us to gain insight. Problems that are difficult for a novice may be easy for an expert, because the latter has so much knowledge and experience that they can recognize a familiar pattern. We have also had our first encounter with dual-process theory and can see that fast, intuitive processes can be both a source of error and a cause of success, depending on the context and the prior knowledge of the problem solver.


The mechanic’s more sophisticated mental model allows him or her to test a series of ever more specific hypotheses, converging efficiently on the cause of the problem.


The purpose of science is not simply to gather facts but to advance knowledge. Science depends upon having good theories. Such theories allow us to understand and predict the natural world and are also the basis for technological advancement.


Popper recommend that we construct scientific theories to be falsifiable and then set about trying to disprove them.


People may be biased and irrational in their reasoning. In particular, he suggested that people were bad Popperians. They seemed to test their hypotheses by trying to make them true rather than false. This tendency has become known generally as confirmation bias.


If is the word we use in everyday language to express hypotheses and to stimulate hypothetical thinking.


I should like to mention our extraordinary tendency to engage in counterfactual thinking. That is, thinking about how things might have worked out differently if something had been changed in the past. “I would have had a great score today if I had holed some putts.”


We also use counterfactual statements to undo past events in our minds, especially when we regret an outcome.


Automatic Type 1 processes dominate most of our everyday choices. The psychology of decision making mostly involves novel choice problems that require Type 2, deliberate thinking for their solution. Errors, however, are often attributed to Type 1 processing.


When we give someone a decision to make, we can get a different answer according to how we ask the question, a phenomenon known as framing.


For some reason, people feel more responsible for actions than inactions and as a result can do harm by their failure to act, a phenomenon known as omission bias.


We can come to know things directly or by reasoning.


What the theorem captures is the idea that our belief in hypotheses after examining some evidence should be determined both by our prior belief in them and the evidence examined. Belief always changes in the direction of the evidence but by variable amounts. This is known as Bayesian reasoning.


Bayesianism also provides an alternative philosophy of science to the logical method of Karl Popper. Bayesians do not seek absolute confirmation or refutation but rather revise their beliefs continually in the face of evidence.


In the new paradigm, conditional statements are given a probabilistic interpretation. Instead of asking people if conditionals are true or false, we can ask them instead to judge the extent to which they believe them, by assigning a probability.


Over time it became clear that abstract, logical reasoning is not something that ordinary people can do at all well without training and that the natural mode of reasoning is belief-based. People also lack the ability to reason correctly with probabilities presented in formal problems, again without special training.


Behaviorists believed that studying animal learning would tell us much about human behavior. The focus was entirely on instrumental rationality (e.g. how animals learn to obtain food) and mostly on basic goals. The behaviorist movement showed scant interest in the representation and processing of knowledge in the human mind. Philosophers, however, had long recognized its importance, discussing epistemic rationality. We are epistemically rational when we acquire true beliefs about the world around us. But we also need to be able to deduce the correct inferences from these beliefs.


Their argument is that behavior is rational because it has been shaped by evolution to be so. However, the environment in which we evolved is not that which we now live.


In fact, he has argued that we are often better off relying on intuitions or “gut feelings” rather than attempting to solve problems by conscious rule-based reasoning.


The RQ (rational quotient) might predict good performance for scientists, engineers, and economists, although even here it might not capture qualities of imagination and intuition that can be important for creativity. It seems less relevant to subjects such as music, literature, and arts generally where an intuitive style would be more productive. Even in fields such as management and politics, success does not just depend upon reasoning but requires good judgment.


The great mass of brain work must be automatic because consciousness has such a limited capacity. It is a commonplace observation that we perform many of our tasks on “automatic pilot.”


While our conscious attention is elsewhere, some process in our brain is still monitoring the road situation for hazards and is able to call up conscious attention when required. Second, it shows us that our ability to think consciously is limited and we need to switch it away from one demanding task to perform another.


Many cognitive psychologists avoid talking about consciousness at all, believing it to be a poorly defined concept. Thus many different terms have been used to describe Type 1 and Type 2 thinking. Some author talk of intuitive and deliberative processing, others of automatic and controlled processing, and still others distinguish processing that is associative from that which is rule-based. Some contrast thinking which is heuristic with that which is systematic or analytic.


It would be a mistake, however, to believe that Type 2 thinking is necessarily superior. It has special properties, to be sure, but it also has a very limited capacity and is relatively slow. The brain must carry out the vast majority of its processing automatically, rapidly, and outside of conscious awareness. Most of this processing is effective and helps us achieve our goals.


Expert problem solving and decision making often arises from fast pattern recognition processes rather than slow reflective reasoning. Experts, like everyone else, have 2 different kinds of knowledge. When training in medical school, they acquire a great deal of explicit knowledge through book learning. They will need to call on this early in their careers and also later when presented with unusual cases. However, when a doctor practice medicine, they also acquire a great deal of implicit knowledge by experiential learning. An experienced general practitioner will not need to recall textbook learning when confronted with a common set of symptoms. In the same way, an experienced police detective will usually know when a suspect is lying, but they did not learn this from a lecture or a book.


These experts are responding to patterns, which may be quite complex sets of cues whose meaning has been acquired by experience. The learning was implicit, the knowledge is implicit, and so the expert experiences what we call an intuition. That is a feeling, without a conscious rationale to support it. This is something quite different from learning explicit rules and applying them consciously by Type 2 reasoning.


While most decisions are made on the basis of habit and experience, Type 2 thinking is required when we need to imagine and compare future consequences of our actions. Such decisions tax our powers of Type 2 thinking to the limit. This thinking appears to be slow, effortful, detached, and can essentially deal with only one thing at a time.


He has pointed out that if people are to solve one of these problems they must first perceive a need to reason. He suggests that we are by nature “cognitive misers” who only expend effort on reasoning when we are well motivated to do so. Intuitive answers can be so compelling that we seem never to engage in reasoning at all.


Animals also have a form of working memory and controlled attention. But clearly powers of hypothetical thinking are far better developed in humans.


Old mind learning and conditioning involve repeating what has worked before, as is typical of animal cognition. But the new mind looks to the future and tries to make decisions by mental simulations — imagining and reasoning about the future consequences of our actions.