Choices: An Introduction to Decision Theory

by Michael D. Resnik

Book by Resnik, Michael

Genres: Philosophy, Nonfiction, Textbooks, Psychology

240 pages, Paperback

First published January 1, 1987

Reviews

[path · Rating 3 out of 5]

It is probably the case that this book has greater overall application for readers interested in machine learning and similar applications. I say this because there is a point in the book where the author goes past the intuitive utility of key concepts and maxims for decision making into the realm of what might best be described as quantitative or algorithmic decision making. Resnick does attempt to illustrate the more algorithmic decision making with realistic scenarios, but for the algorithms to work, the decision scientist must reduce some very messy and subjective qualities into numbers. This is mostly fine when we are talking about measuring preferences or regrets because we are probably accustomed to thinking about them on a scale (“on a scale of 1 to 10, how do you feel about …”). But when we need to start taking into account things like utility and the influence of desire and emotion and the intensity of experience, the conversion of those qualities to numbers seems a lot more problematic. And the more Resnick leans into this quantified approach to decision making, the more it seems that a) no human can engage in this level of decision making without computational support, and b) relying on that computational support can lead to some potentially impolitic, rude, and insensitive decisions — rational though they may be. To be fair, though, Resnick acknowledges this difference from the outset, when he says that there is “descriptive decision theory” which concerns itself with how decisions are made (or modeled, I’d say) and “normative decision theory” which concerns itself with how decisions ought to be made (3).

Overall, I would say that Chapters 1 - 3 mostly stay in the realm of lived, relatable experience. Chapter 4 (on utility) starts there but gets a little too intricate and mathematical to be traced in a realistic way to lived decision situations. Chapter 5 (Game Theory) and Chapter 6 (Social Choices) are interesting and well articulated, and it is clear that we face decisions all the time that might be modeled by game theory or are indeed subject to social and group influence. At least conceptually the chapters are valuable to decision makers in that they show how the rules of games and subjective and interpersonal elements of groups influence choice. Whether these concepts can be applied with the same kind of rigor and precision as decisions are presented in the early chapters is up for debate. At the very least I would say that Resnick does offer some insights about the factors that can influence decisions and gives a sense about how, as variables, they would influence decisions. So, I’m not disappointed. I still think that I learned something about how to think about decisions and how to weigh, at least qualitatively, different factors.

Some Takeaways

Resnick invites us to think about decisions as involving “acts,” the choice options, “states,” or things that are true or false about the world regardless of our choices, and “outcomes,” which are what follow from the intersection of our choices with states in the world. We can present these as a matrix with acts on the rows and states in the columns and each cell represents a different outcome (7). They can also be represented as branching decision trees with boxes representing decisions and circles representing chance that certain outcomes will follow (17). The challenge in setting up these matrices or trees is in “problem specification” (8) or thinking through the relevant states, choices of acts, and likely outcomes. In my experience, this is something that people are often not good at or don’t spend enough time doing and so wind up with a very limited or biased sense of the choices and outcomes that we are trying to make.

Resnick further specifies that we make decisions under a couple of conditions:
* Certainty - where we know what the desired outcome is an how to make a choice that brings it about
* Risk - where we cannot make a certain choice but we can assign probabilities that certain outcomes will happen
* Ignorance - where we cannot make a certain choice or assign probabilities to any outcomes.

Chapters 2 - 4 are dedicated to showing how to model decisions under risk and ignorance because decisions under certainty really require no explanation. Some useful concepts that arise from these discussions are:
* Orders of preference that logically specify all of the possible preference orderings: e.g., if X is preferred to Y then Y is not preferred to X. And if X is preferred to Y then X is not indifferent to Y. And if X is indifferent to Y then neither X is preferred to Y nor is Y preferred to X … etc. (23)
* Maximin Rule: when ordering preferences by preference or utility choose the act that has the highest (max) minimum (min) value. (26)
* Minimax Rule: when ordering likelihood of regret for not making a decision, minimize (mini) the maximum (max) regret (28)
* Optimism-Pessimism Rule: a formula for thinking about how to assign probabilities of preference or regrets (33).

These articulations and logical constructs are what set up the basis of the algorithmic decision making that follows, but I won’t belabor it (any further). Overall a worthwhile read. Conceptually, some of this will stick with me and I think that my interest will be satisfied more by looking at how others have taken up work like this into more intuitive, lived decision making scenarios.

[Chris Meyers · Rating 3 out of 5]

If you want to acquire a basic understanding of decision theory (and who doesn’t?), then this book might interest you. The book covers all the basics of decisions under ignorance, decisions under risk, and game theory. The author also does a good job of explaining the relevance of the principles, the practical application of the theory, and the philosophical problems underlying the disagreements and paradoxes.

The book is aimed at novices, and is intended to be helpful to people with varying interests—philosophy, social science, economics, etc. Thus it does not require math skills beyond college algebra. Despite being an introduction, the book covers a lot of ground.

The major drawback of the book is that often it covers too much ground too quickly. The formulas get very long and complicated rather quickly; and the explanations are sometimes lacking. The reader is expected to work through some of the derivations for herself, which is very difficult if (like me) you have not done any algebra in a while. It would have been better if the book were a bit longer and moved a little slower. Each section is followed by a several problems for the reader to work through, but there are no answers in the book (so if you don’t understand how to work the problems, then they are useless).

In the end I learned a lot about decision theory, but I feel that I will need to read another book on decision theory to fully grasp what I should have learned in this book.

[Adam · Rating 3 out of 5]

Very readable, with really interesting exercises that aren't too fiendish nor too easy.

[Alexander Barnstone · Rating 4 out of 5]

A great book for introductory normative decision theory. Resnick gives a comprehensive and balanced look into many different facets of the theory, from paradoxes of utility to game theory. There are some grammatical issues that could be fixed, and at moments better explanations are warranted. But overall very good.

[Mitch Flitcroft · Rating 4 out of 5]

This is a good introduction to the field of decision theory. It covers decisions under ignorance, risk, and certainty, game theory, and social choice theory. The book doesn’t require a background in mathematics, although at least some knowledge of propositional logic is helpful.

[Ron Me]

Most of the book was rather standard game theory with a little philosophy tossed in. But the last 15% had things about multi-player game theory and Social Choice that I hadn't seen before. So yeah, I recommend it.

[Brendan McNabb · Rating 3 out of 5]

I picked up "Choices" thinking it would accessible, and more abstract. It is not. The book is probably very good, if you're taking a class where you need to to understand decision theory. If I understood this book more, I would rate it much higher. Trying to read this book thinking it would relate to interpersonal choices, more abstract game theory, it was a mistake on my part thinking that's what I would get out of this text.