The question of how the brain divides up a scene in order to interpret it is a controversial one. Some argue that it sorts visual inputs according to the nature of the objects present, others hold that divisions are made according to where things are in the visual field, and others still put partitioning down to movement.
According to new research published in Nature1 (7 October 1999), the brain does all of the above. Just as when you highlight piece of text on the computer screen your mouse selects not just the text but its font, type size, line spacing and so on, the brain automatically 'selects' all the visual attributes of the object it is attending to, whether they are relevant to the task in hand or not.
In other words, even when we are nominally paying attention to one visual attribute of an object such as a moving face – 'it is moving', or 'it is a face', for instance – the brain processes 'it is moving and it is a face'. As Nancy Kanwisher of the Massachusetts General Hospital, Charleston, Massachusetts, and colleagues comment in their paper, "our data provide the first neural evidence that objects serve as the units of attention even when the selection of objects is not required by the task".
One of the images used in the experiment. In this example, the face was moving rapidly from side to side.
Kanwisher's group made this discovery by using functional magnetic resonance imaging to monitor neuronal activity in three disparate brain areas, the fusiform face area, the parahippocampal place area and 'area MT/MST' – regions that respond most strongly to faces, places and motion, respectively. They showed test subjects images consisting of two pictures layered in the same spot – such as a face transparently superimposed upon a building – in order to preclude location-based selection. They then presented their volunteers with random trials wherein one layer – either the house or the face – oscillated very rapidly while the other remained stationary and asked them to monitor the faces, the houses or the direction of motion, fixating on a central spot all the while.
This revealed, as expected, that the magnetic resonance signal in a brain region, the fusiform face area, say, was greater when subjects attended to the attribute pertinent to that region, in this case, the face. But these experiments also turned up a more unexpected result. When the brain is paying attention only to 'faces', say, it also seems to register irrelevant attributes – movement, in this example – even if the two attributes are effectively independent. That is to say the magnetic resonance signal in the 'area MT/MST' went up when the face was moving, for example, even when the task was 'attend to faces'.
"It is unlikely, however, that objects are the only units of attentional selection," Kanwisher's team concludes. The roles played by other types of selection in working memory and perceptual decision making remain to be seen.
O'Craven, K.M, Downing, P.E. & Kanwisher, N.fMRI evidence for objects as the units of attentional selectionNature401, 5841999.