Origins of System 3
The idea that our thinking operates in two distinct modes, one fast and intuitive, the other slow and deliberate, emerged from the lifelong work of two Israeli psychologists, Amos Taversky and Daniel Kahneman. Kahneman immortalized their insights in his book, Thinking Fast and Slow (2011). This work both changed how many people thought about and understood how decision-making and judgment actually work, and laid important intellectual foundations for the emergence of behavioral economics.
My own PhD work was focused on questions of what we can know and how, an area of philosophy called epistemology. I became part of a small group of people at University of Toronto (all of us working under the supervision of professor Ian Hacking) who worked on what Hacking started by calling “funny studies” and ended up branding, historical epistemology.
I was exploring communities of practice in law and science who often used the word objectivity, but in different ways. I developed a hypothesis that while ideas of objectivity are not only important, but critical to some of the most consequential work in these fields, objectivity isn’t what many people seem to think it is. Far from there being a single and universal idea or test for objectivity, there are, I argued, many different kinds.
I became particularly focused on a key idea of objectivity in the law of self-defense, where the question of objectivity is centered on a test known as the “reasonable man standard.” This idea emerged from British Common Law jurisprudence (“codified” as precedents in the decisions of judges, as opposed to statutes passed by legislatures). The concept of dates back to the late 16th century and came to guide decisions about what we ought to expect a reasonable person to behave in a given set of circumstances.
In my dissertation, Reasonable Men and Battered Women: Objectivity in the Law of Self-Defense, I attempted to make the case that feminist lawyers had exposed the fact that the “objectivity” of the reasonable man standard was misapplied to women in general, and abused women in particular, because it depended on assumptions about what a “reasonable” person might do in a physical confrontation between two people (all the historical cases referred to altercations between men) of relatively equal size and strength. Not only was there a historical male bias that had entered through case law, but that bias was systematically disadvantaging women.
The purpose of the objective test in the law of self-defense, the feminist lawyers argued, was to help juries decide what was reasonable in the context. In the case of women who are abused, they argued, what is reasonable is shaped by their experiences with their violent partners and the beliefs that have developed through repeated experiences of abuse. As a result of a number of cases in the UK, the United States and Canada, the law and the understanding of the application of the standard began to change.
What I came to notice through the study of these cases in the law of self-defense, and in the case of scientific paradigms that change over time, is that there are constant conversations among those in professional communities that are themselves part of the practice of both law and science. These kinds of conversations, especially in communities of specialized practice, have fascinated me for my whole life.
In the Spring of 2022, after nearly 30 years of working in design, strategy and innovation communities, I started exploring ideas about conversation and intelligence with a group of people from around the world. The more conversations we had, the more I began to wonder whether it might be possible to build a machine, something we might now call an artificial intelligence, that could participate in human conversations. Convinced that this might be possible, I began to think about starting a venture to build such a system.
With enough of a background in cognitive science, computer science and the history of ideas of intelligence to be dangerous, I started to articulate a concept sketch vision of such a system. I called it “Our Teacher” and represented its mediating intelligence with an image of an octopus, in homage to the film, My Octopus Teacher. Determined that the project have a practical application and a business case, I decided that the meetings people have at work was a sufficiently large and sufficiently well defined context to form a problem set that could be addressed by a venture-backed startup.
Alas, this one never got out of the garage. But even after I decided to let go of the idea of the startup, I couldn’t let got of the ideas nor the concerns that had motivated me to explore then in the first place.
The ability to have effective conversations at work, from the mundane, everyday ones that we use to start, guide, monitor and evaluate projects, to the conversations in which we debate and decide on strategy, discuss and form policy, has never been more important. We operate in environments of escalating complexity and in societies of declining civility. The most important and consequential conversations we need to have, how to confront and end systemic discrimination, how to prevent social and planetary collapse, what to do if we cannot prevent such collapse, are just some of the conversations we are either not having at all or doing very poorly at.
The need to have these conversations, however, is unavoidable, I think. There are two big reasons for this. First, we are social creatures and we have really only two methods for making decisions: one is through agreement and the other is by the use of force. We will not be able to force people to take the actions necessary to solve the complex problems we most need to confront. Yet, we also have reason to doubt whether we can reach agreements in an increasingly uncivil world. Second, it is in conversation, and there alone, that we can access a form of intelligence and a way of thinking that is urgently needed to navigate complex. problems.
These are the reasons I came to believe that in addition to the kinds of thinking that Taversky and Kahneman had so powerfully described, there is another kind of thinking that stands outside of their account. So, while I think that the explanatory power of thinking fast and slow has been immensely valuable and acknowledge that their ideas about System 1 & 2 has unlocked entirely new fields of study and understanding, I also think we need a System 3.
I am not only interested in System 3 because I believe that it will may us better understand a kind intelligence and a way of thinking that I think we urgently need a better command of, but also because I think that it might also lead us to better understand why diversity matters and how it can lead us to a better model of performance in the places we all spend most of our waking time. At work.