Noyce Conference Room
Seminar
  US Mountain Time
Speaker: 
Kelle Dhein (University of Kentucky)

Tune in for the live stream on YouTube or Twitter.

Abstract: Biologists regularly apply the concept of information in their work (e.g. echolocation gives bats information about their surroundings, DNA carries information about how to build an organism), and while their usage makes intuitive sense in context, it is difficult to articulate precisely what they mean when they invoke information. Some argue that Shannon’s technical, mathematically defined notion of information is the proper starting point for understanding informational content in biology. Others have sought to define content-presupposing concepts like information in terms of the historical selection processes that drive evolution. However, it is unclear whether these approaches have produced definitions that capture the way successful researchers in the behavioral sciences use content presupposing concepts like information. In this talk, I emphasize the importance of attending to scientists’ investigative practices and historical context when confronting problems about what scientists mean when they invoke the concept of information.
 
First, I examine a successful, longstanding experimental tradition whose practitioners have consistently ascribed informational content to the neurosensory mechanisms of insects. By analyzing the way scientists design and interpret experiments to justify claims about information, I clarify the norms guiding scientists’ ascriptions of content to produce a pragmatic definition of information. I argue that the norms guiding scientists’ ascriptions of content bear some resemblance to formal information theory, that they rely upon a cybernetic notion of biological function, and that they have practical value for scientists.
 
Second, I adjudicate the cognitive map debate in animal navigation studies. Mammalian navigation researchers have long held the cognitive map hypothesis to be an established fact. However, the cognitive map hypothesis has engendered an ongoing, decades-long debate among insect navigation researchers, especially those referenced in the last paragraph. By attending to the personal histories of the scientists leading the debate and the broader history of animal behavior research, I show how older debates about instinct vs. learning that were supposedly neutralized in the 20th century continue to motivate this debate over how insects process information.
 
Finally, I ask a question that encompasses both problems addressed so far: What ever happened to cybernetics? In the late 1940s, the information age was most clearly heralded by the newly mobilized science of cybernetics, which promised to synthesize human, animal, and machine behavior using recently formalized notions of information, feedback, and control.  As the 21st century drew nearer, most academics came to regard cybernetics as a dead movement. Nevertheless, some behavioral scientists continue to see revolutionary potential in the concept of information for synthesizing understandings of behavior, and although researchers continue to grapple with the problem of comparing animal and machine behavior, biologically-inspired advances in artificial intelligence seem to be realizing the cyberneticians’ prophesy. For example, the insect navigation research referenced in the last two paragraphs has been used to design autonomous robots. I conclude by  arguing that current attempts to use information to compare human, animal, and machine behavior would benefit from historical and pragmatic contextualizations that reveal how such projects relate to their fore-bearers from the dawn of the information age.

Purpose: 
Research Collaboration
SFI Host: 
Melanie Mitchell