1. "How They Know Where They're Going: New Research On Cabbies Shows That The Brain's Right Hippocampus Is Key For Navigation" -from the "Society for Neuroscience" September 16, 1997

Eleanor Maguire and her team at the London Institute for Neurology recently finished a study of London cabbies and areas of the brain responsible for navigation memory and recall. The research team studied PET (positron emission tomography) images of the taxi drivers' brains when asked to recall the route to a specific destination. PET images were also studied from asking the same drivers to recall plots of films and famous landmarks. The results from this study showed that to recall sequential events, such as plots, the left frontal lobe was stimulated, and in landmark recall, several areas were activated, including the occipitotemporal regions, posterior cingulate gyrus, medial parietal area, and parahippocampal gyrus. Navigation memory used all these areas, but also involved the right hippocampus, which was only activated during this specific recall type.

Issues in computational intelligence raised by this: What levels of explanation are invoked by this problem? What kinds of representation must be stored to account for these results? How does the debate between modularity and Interactionism come into the picture?

2. "Artificial Intelligence Improves Heart Attack Diagnosis" -from "AHA News Media Relations" September 16, 1997

Drs. Lars Edenbrandt and Bo Heden have recently reported that an artificial neural network designed to diagnose heart attacks has shown to be 10% more effective in identifying abnormal EKGs (electrocardiograms) than experienced cardiologists on staff. This neural network was trained using 1,120 EKGs of people who had heart attacks and 10,452 normal EKGs. It is currently estimated that over 25% of EKGs are misdiagnosed by physicians. Therefore the new neural network will be used as a tool, but will not take the place of a doctor. Edenbrandt is quoted as stating, "the EKG is only one of several tests used by physicians to diagnose a heart attack. Doctors will still need to talk to patients about their symptoms and medical history."

Issues in computational intelligence raised by this: What role does modeling play in cognitive science? More particularly, what is the relation between theoretical and applied cognitive science? What role do expert systems have in specialized pursuits? How does this underscore the prospects and limits of neural nets (as learning devices, trained on specific input over many cycles)?