- Amazon.co.jp ・電子書籍 (449ページ)
Thinking, Fast and Slowの感想・レビュー・書評
人の頭には素早く自動的に考える「Sytem 1」とゆっくり慎重に考える「System 2」があるということをベースに，人の考えの癖について様々に論じた本．無理して英語版を買ってしまい，読了までに 1 年もかかってしまった :-) 前半の内容はほぼ忘れてしまっているが，必ずしも「System 1」が悪く，「System 2」が良いというわけでもない．両者の特徴を理解した上で，正しい判断をしていくことが重要である．
An amazingly well-written book by the laureate of the Nobel Memorial Prize in Economic Sciences 2002. In this famous book, an Israeli-American psychologist explains his study in psychology and behavioral economics. Written in a casual way, he uses the style of essays and talks not only about academic studies, but much about his life and his colleagues. One can read this book as his autobiography.
What is striking for me is failures and mistakes in his many experiments. He writes much on how to plan an psychological experiment, and makes us recognize how much findings come from its failures.
Main theme of this book is to analyze the rational-agent model of the classical economics and to investigate real mechanism, propensity and character of our cognitive system. The classical economics describe economical agent as being coherent, logical and rational (which Kahneman puts `Econs'), which does not always hold on us as a psychological agent (which he puts `Humans'). That is, "[t]he Econs of the rational-agent model do not resort to mental accounting: they have a comprehensive view of outcomes and are driven by external incentives. For Humans, mental accounts are a form of narrow framing; they keep things under control and manageable by a finite mind"(loc.5901). By saying this, Kahneman does not assert Humans ares irrational, but simply they cannot be explained by the rational-agent model as Econs(loc.7096).
This means we need other factor to figure out our way of thinking. An apparently irrational judgment is not due to the corruption of thought by emotion. As revealed in many experiments, there exist systematic errors in the thinking of normal people. We need to investigate another machinery of our cognitive system(loc.172).
So, what should be a model of Humans? This leads to the title of this book: Thinking, Fast and Slow. In short, Kahneman distinguishes two characters in our psychological cognition -- called System 1 and 2, proposed by psychologists Keith Stanovich and Richard West(loc.324). "The two characters were the intuitive System 1, which does the fast thinking, and the effortful and slower System 2, which does the slow thinking, monitors System 1, and maintains control as best it can within its limited resources"(loc.7037).
Much effort of this book goes to clarify the character of System 1. It produces impressions, intuitions, feelings and others, but this work in our mind without notice(loc.91). The most remarkable trait is its autonomy and speed: System 1 operates automatically and quickly, with little or no effort and no sense of voluntary control(loc.324). We usually judge and act guided by products of System 1, and in most cases, it occurs no problem(loc.93,392). It is sure System 2 has ability to override and correct products of System 1, but there is no need to activate its ability -- in most cases, but not always.
What is the raison-d'être of System 1? Kahneman points out that "[t]he capabilities of System 1 include innate skills that we share with other animals"(loc.344). These innate skills are old in view of evolution, and one might say these lies somewhere in the limbic system of our brain. System 1 is there to "provide a continuous assessment of the main problems that an organism must solve to survive: How are things going?"(loc.1543). In short, System 1 represents what is normal and what is not so that maintains and updates a model of one's personal world(loc.1222). It must be well noticed that System 1 does not only represent a status of environment in its present time, but links the present with recent past and near future (expectations). System 1 compares the current status with the past one, and determine whether it is normal or surprising (and there is requirement to act differently than usual). All these go under consciousness, without notice.
This function of System 1 is quite familiar with philosophers, especially phenomenologists. At all, this book may be regarded as today's version of Merleau-Ponty's `Phenomenologie de la Perception'. The silent process of System 1 is well discussed under the concept of `Passive Synthesis'(Husserl), `Immer Schon'(Heidegger), `Original Oui'(Derrida), etc. The construction of a model of environment is also well known as Umwelt(Uexküll) or broadly expandable to Seinsverstandnis(Heidegger), and the significance of past and future in model construction recalls analysis of Zeitbewußtsein by Husserl. A model of environment System 1 provides is always starting point of our action. We cannot turn System 1 off(loc.410). This world is always at first presented by System 1 -- aletheia or Erschlossenheit.
On building up a model of environment, there are strong limitations for System 1: scarcity of information and timeliness. For System 1, available information at one time is fragmentary. By linking them with its disposal, it makes them to a coherent causal theory(loc.1297). So there always remain gap and jump between given information and conclusion of System 1. Moreover, what is important is: "it is not designed to know the size of its jumps"(loc.3561). This means that System 1 always presupposes stable regularities in the environment(loc.4142). This recalls `Transzendentale Schematismus der reinen Verstandesbegriffe' of Kant to me. Speaking of timeliness, there is no time for System 1 to doubt and criticize the given information. Taking long time to grasp the surround situation poses serious risks -- for example, one cannot run from high-speed predators. Therefore, "System 1 is not prone to doubt. It suppresses ambiguity and spontaneously constructs stories that are as coherent as possible"(loc.1911). All information System 1 can use is ideas currently activated. it cannot allow for information it does not have(loc.1471) (But to notice, it uses information from memories).
Kahneman creates an abbreviation for this character of System 1: WYSIATI(what you see is all there is)(loc.1486). System 1 is prone to accept the given information regardless of what we want to know. Therefore, it is easily influenced by existing information: Priming effect(loc.2063-2167). Other characteristic exposed by experiments is substitution(loc.1674): we often substitute difficult question with easier one(intuitive heuristic; loc.257) and the answer to an easy question (How do I feel about it?) serves as an answer to a much harder question (What do I think about it?) (affect heuristic; loc.2352). In this respect, a coherent story System 1 provides is plausible, not probable. It is not made in view of logical truth, but of heuristic truth. This is why System 1 often seems to break logical laws(loc.2719). Speaking of memory, System 1 does not represent the exact past event. It evolved to represent most intense scene of the episode and the feelings when the episode was end(loc.6615-6685;7039-7061). In summary, "[t]he world in our heads is not a precise replica of reality; our expectations about the frequency of events are distorted by the prevalence and emotional intensity of the messages to which we are exposed".(loc.2344)
In contrast with statistical facts, more lights cast on character of System 1. "There is a deep gap between our thinking about statistics and our thinking about individual cases. Statistical results with a causal interpretation have a stronger effect on our thinking than noncausal information. But even compelling causal statistics will not change long-held beliefs or beliefs rooted in personal experience"(loc.2997). Regression to mean is a good example(loc.2997-3081,7573). System 1 is not easily educated(loc.7206). (Regression to mean is also strange for System 2.; loc.3368)
A great remark on System 1: Overconfidence. "[M]any people are overconfident, prone to place too much faith in their intuitions"(loc.748). Of course, these result System 1 provides may be modified by System 2 using attention and memory(loc.366,602). But System 2 is mostly constrained to information that is consistent with existing beliefs, not with an intention to examine them. Sometimes System 2 becomes apologist for the emotions of System 1 rather than critics(loc.1779). Moreover, we do not notice of this overconfidence -- we are blind to our blindness(loc.389). This overconfidence and second-order blindness have a important role in our cognition: it makes easier to construct a coherent story of environment, paradoxically(loc.3424,4382). Our almost unlimited ability to ignore our ignorance enables us to make this world sense for us. (This will be a hint for the famous `frame problem' in artificial intelligence.)
Now, speaking of System 2. "System 2 is activated when an event is detected that violates the model of the world that System 1 maintains"(loc.399). It has tasks of self-control(loc.430). There is a division of labor between System 1 and 2, which is highly efficient(loc.405) in sense that System 2 works only when something goes wrong with results of System 1, otherwise System 2 just accepts them.
Interestingly, Kahneman characterizes System 2 as lazy. "Laziness is built deep into our nature"(loc.583). "The laziness of System 2 is an important fact of life"(loc.2813). Laziness of System 2 comes from fact that the exertion of self-control is depleting and unpleasant(loc.701). Doubting requires a lot of mental effort because it is to examine several inconsistent interpretations at one time(loc.1376,1913,6134). So System 2 often tries not to work, just accepting results of System 1. (This laziness of System 2 varies from one person to another; loc.815)
Based on results of analysis as System 1 and 2, Kahneman puts: "[t]he best we can do is a compromise: learn to recognize situations in which mistakes are likely and try harder to avoid significant mistakes when the stakes are high"(loc.461). One of valuable points of this book is to provide many ways to avoid (or rather utilize, even abuse) cognitive errors. Basically, the strategy for avoiding cognitive errors is to slow cognition down and ask for reinforcement of System 2(loc.7211). This book contains many other example for application. 4 factors For writing persuasive messages: High-quality paper, bright colors, and rhyming or simple language(loc.1091). Psychology can contribute to democratic politics as providing strategy for managing people's emotion(loc.2467). However it seems paternalistic, we as Humans need some help for making good decision(loc.7099-7164). A very famous book on management, Jim Collins and Jerry I. Porras's "Built to Last" is glorified by the halo effect and outcome bias(loc.3530). There is a great difference between experiencing self and remembering self and interests of these two selves need not coincide, so we must consider in two aspects when we think about happy and good life(loc.6718-6925,7073,7995-8011)
Prospect Theory, as well known as his name, comes with analysis of System 1. This theory is contrasted with Bernoullian utility theory. But unlike it, Kahneman developed prospect theory as a psychological theory that would describe the choices people make, regardless of whether they are rational(loc.5389). It differs from utility theory in the relationship it suggests between probability and decision weight(loc.5593). More concretely, "[t]he fundamental ideas of prospect theory are that reference points exist, and that losses loom larger than corresponding gains"(loc.5085;5085-5229). Among them, the reference points are most important variable in prospect theory, which Bernoullian utility theory lacks(loc.4799,4809). These character of prospect theory have their back in three cognitive features: adaption level, principle of diminishing sensitivity, loss aversion(loc.4815-4829). But surely prospect theory is not perfect: it cannot deal with disappointment and regret (of course Bernoullian utility theory cannot). In case of disappointment, the value of an outcome to change when it is highly unlikely, or when the alternative is very valuable(loc.4915). And in case of regret, it is not the case that available options in a choice are evaluated separately and independently, and that the option with the highest value is selected(loc.4918). But I have felt these limitations of prospect theory are not well explained in this book.
Finally, but perhaps most important point of this book: "System 1 and System 2 are so central to the story I tell in this book that I must make it absolutely clear that they are fictitious characters. Systems 1 and 2 are not systems in the standard sense of entities with interacting aspects or parts. And there is no one part of the brain that either of the systems would call home"(loc.476,7168). Maybe we can guess that System 1 lies somewhere in limbic system, System2 lies somewhere in cerebral neocortex or frontal lobe. But this is complete another problem (and for another person, like neuroscientist).
Thinking, Fast and Slowのその他の作品
Thinking, Fast and Slowを本棚に登録しているひと