Until very recently, humans were much better at face recognition than any computer vision system. But all of a sudden computers seem to be nearly matching human performance. But might we still be better than computers in more real-world face recognition tasks?
In some situations, humans are surprisingly bad at face recognition.
People differ markedly from each other in their face recognition ability. Face recognition ability is heritable and is not correlated with IQ, and some otherwise normal, perfectly smart people are so bad at face recognition they routinely fail to recognize close friends and family members.
This talk uses face perception as a case study to illustrate the power of low-tech behavioral methods; observations from reaction and time and accuracy in face perception tasks reveal “signatures” of face recognition (inversion effects, composite effects, part-whole effects) that yield fundamental insights about how we...
It was once thought that face recognition takes abut ten years to develop fully, but more recent research shows that adult-like face recognition is present at the earliest ages scientists have been able to test it.
Experience with faces does affect our face recognition abilities, but not in the way you might expect.
Here I describe the basic fMRI evidence for the fusiform face area, how we test alternative hypotheses to face specificity, and the functional region of interest method.
In this astonishing video from Josef Parvizi and Kalanit Grill-Spector at Stanford, you meet a man who has electrodes right on his face area (for medical reasons), and he tells us what he sees when that region of his brain is stimulated.