July 14, 2014
Stanford scholar takes a philosophical approach to human behavior
By analyzing how people cooperate and make plans, philosophy Professor Michael Bratman creates a framework for understanding human sociality that has implications in fields ranging from psychology to artificial intelligence.
By Marguerite Rigoglioso
What's the difference between walking down Fifth Avenue alongside strangers, and purposefully walking down Fifth Avenue with someone?
It's the kind of philosophical question that gets at the heart of what may separate human from ape, yet no one has come up with an adequate answer.
Given the importance of shared activity to the survival of the human organism, any scientific way of thinking about the mind and human action, or "agency," needs to be concerned with it. It may be surprising, then, that little has been written about the topic in fields such as cognitive science, psychology and philosophy until relatively recently.
New research by philosopher Michael Bratman, the U. G. and Abbie Birch Durfee Professor at Stanford, is therefore poised to make an important contribution to numerous fields by offering a new model for understanding basic forms of human social interaction.
An expert in philosophy of action, Bratman details his findings in his latest publication, Shared Agency: A Planning Theory of Acting Together (Oxford University Press, 2014).
So, from the conceptual standpoint, just what distinguishes walking alongside someone from walking together?
Bratman's main claim is that the key ingredients in human shared activity are the intention to do something together, combined with the interlocking and intended meshing of plans – my plans that we walk down Fifth Avenue together are connected and mesh with your plans that we do so.
"Shared intentional activity is a complicated structure of interconnected people who are planning," asserts Bratman, who has also written widely about practical reason, free will and moral responsibility. The bulk of his new research is devoted to describing that structure – the interpersonal connections, meshing and responsiveness that go on in small group interactions.
By reflecting deeply on how humans engage with one another, Bratman brings to light the fact that people have a natural ability to create "unity amidst diversity" – which has important implications in the arena of politics and civic activity.
"People frequently join forces on things without having to agree all the way up the line. It's something our current political system could do well to work with more skillfully," he said. "Reasons why may not be as important as the intentions and agreement to work together on some cause or issue."
As his theory illustrates, people can have a shared intention – say, to paint a house together – even if their reasons are quite different. "You may want to paint because you don't like the original color, while I want to do it because I can't stand the mildew," he said. "Neither of us may care very much about the other's reason, but we still share an intention to paint, and we thus organize and unify our activity downstream."
Bratman's theory is also relevant to other arenas concerned with making group activity more efficient, including business and organizational behavior, and his work on shared agency has been influential in the field of psychology. His theoretical work affirming the importance of acting together as a fundamental feature of human social psychology extends even to the arena of primatology.
"Why do humans create operas and build buildings, and great apes don't – even though they are very close to us genetically?" Bratman asks, drawing on work of the developmental psychologist Michael Tomasello.
Tomasello, co-director of the Max Planck Institute for Evolutionary Anthropology in Leipzig, Germany, has found that great apes have planning capability, as evidenced in their ability to hunt in groups. But, as Bratman notes, "According to Tomasello, they're acting strategically, yet not together – kind of like people walking down Fifth Avenue beside one another but not together." Bratman's work provides a way of articulating Tomasello's striking conjecture that what the apes lack is the capacity for shared intentionality.
A common ground of shared commitments
Although the Stanford theorist's new book focuses on "duets and quartets" – that is, smaller-scale interactions – he says it also may be useful in the great debate about holding organizations responsible.
"Part of that debate is whether organizations can have intentions in the way people do," he said.
As a research fellow at the Stanford Humanities Center next academic year, one of Bratman's projects will be to explore how his ideas might be applied to other larger-scale social phenomena such as the legal and political systems.
One of the chapters in his book, for example, looks at how shared intention relates to the process of "shared deliberation" – negotiations and discussions over things like what a startup needs to do to ramp up its business activities, who should be hired for a job and so forth.
"Shared deliberation is different from what social scientists call 'bargaining,' or the 'I'll do this if you do that' approach to joint activity,'" Bratman said. In ordinary bargaining, each party brings to the table his or her own concerns without assuming that their issues matter to the others.
As Bratman articulates it, in shared deliberation people reason together in a way that involves a common ground of shared commitments to particular values or standards in their decision making.
Consider, for example, a college admissions committee. Some schools may give greater weight to potential students whose parents have gone to the school, while others may not. Individuals working together in a committee may subsume their personal opinions as to the importance or justness of one policy over another in service to the group standard – which thereby fosters greater cooperation and efficiency.
Other examples include a company's deliberations that are framed by a shared commitment to product safety, or the shared commitments of a legislative body to giving weight to religious tolerance in its deliberations. And Bratman's conjecture is that planning theory can help illuminate these important phenomena.
Groundwork for artificial intelligence
Bratman began to study human agency more than 27 years ago. "I was a young faculty member, a young husband and a young father. My days involved finding ways to balance these different elements in a consistent and coherent way. That meant a lot of planning.
"And it struck me – in part by reflecting on important work of Gilbert Harman – that such plan-infused thinking is a pervasive aspect of our minds and our actions."
Bratman's first book, Intention, Plans and Practical Reason (1987), drew philosophers' attention to the complex roles that planning plays in human activity. His thinking on human agency at the individual level coalesced into what he calls the "planning theory."
"My theory focuses on the way humans are able to achieve quite striking results because we're able to coordinate our activity over time," he said. "When you plant a garden, you must first dig out weeds, fertilize, put seeds in. It only works because of our ability to organize activities temporally.
"It's a commonsense idea, but the phenomenon of humans' ability to plan had not been adequately theorized."
By articulating the structures of the human capacity to plan, Bratman's work became of keen interest to the artificial intelligence community.
Bratman's detailed step-by-step layout of the human planning process gave artificial intelligence programmers a framework on which to build.
In recognition of this, in 2008 the International Foundation for Autonomous Agents and Multiagent Systems gave its highest award to one of Bratman's research papers, co-written 20 years earlier.
"There's an old problem in artificial intelligence about how to make sure that robots didn't first paint the room and then find themselves painted into a corner. That's a breakdown in planning," he said. "You want to make sure that if we have systems like computers and robots that interact with other systems, including humans, that the organization of the shared activity has things going in the right sequence."
For more Stanford experts in philosophy and other topics, visit Stanford Experts.