Attention Parietal Function And Binding

Whether the forgoing account is correct or not, functions that theoretically rely on a spatial map (e.g., serial search for the conjunction of two features) should be dramatically affected by spatial deficits, while those that do not rely on such maps should not (e.g., feature search). I have already discussed the evidence confirming this prediction for conjunction and feature search in neurological patients with spatial deficits. These results are consistent with feature integration theory's proposal that spatial attention is needed to bind features together to detect a target in a multi-item display. Treisman and Schmidt (1982) introduced another test of spatial attention's role in binding. When normal perceivers divert spatial attention and must report the color and shape of items in a 2 or 3 item display, perception of the shapes in the wrong colors occurs with some regularity. The wrong color is most often a color from another item in the display as opposed to a color that is not present but is part of the response set. In fact, confidence in the color of shapes that are perceived can be quite high even when these "illusory conjunctions" occur. Critically, Treisman and Schmidt (1982) found that illusory conjunctions were only observed when attention was diverted and displays were shown for a brief period of time (between 90 to 170 ms for different participants). They argued that under such conditions there was inadequate time for spatial attention to be allocated to the location of all the items, resulting in the miscombination of the shape and color in perceptual awareness. In order to bind the two features appropriately (whether under brief exposure or during conjunction search) spatial attention was involved (Treisman, 1988).

Given the brief exposure duration required to produce illusory conjunctions in normal observers, alternative explanations were suggested, such as memory confusions that could occur between offset of the display and participant report. Such alternatives have been ruled out by results in neurological patients by showing that illusory conjunctions can occur under free viewing conditions. For instance, the Balint's patient RM produced up to 38% illusory conjunctions when presented with only two letters in two different colors for several seconds. His illusory conjunction rate did not change significantly for display times that ranged from 500 ms to 10 seconds (Robertson et al., 1997). Without an adequate spatial map to guide spatial attention, illusory conjunctions were common for RM, and they were consistent with his poor conjunction search performance even in displays of only 4 or 6 items (see discussion in previous section). Such findings have since been replicated in other patients with Balint's syndrome (Humphreys et al., 2000).

The deficits in Balint's syndrome occur when dorsal occipital/parietal lobes of the angular gyrus and dorsal extrastriate areas of both hemispheres are functionally disrupted (see Hof, 1989; Rafal, 2001; Rizzo and Vecera, 2002; Robertson, 2004). RM's lesions were concentrated in these areas with some extension into more superior parietal regions in the left hemisphere.

Neither the calcarine cortex nor ventral occipital or temporal lobes were anatomically damaged (see Friedman-Hill et al., 1995 for reconstruction of his MRI). Consistently, RM's primary vision was intact (e.g., 20/20 in each eye, normal color vision, normal contrast sensitivity, etc.), suggesting that these areas were functionally intact as well.

Some investigators have argued that the deficits in binding and visual search are not a result of disrupting parietal spatial functions per se, but instead are due to the difficulty of conjunction vs feature search and the confusing visual signals that can occur in multi item arrays after brain injury. These arguments have been quieted by evidence acquired with functional imaging measures in groups of normal perceivers. For instance, in an fMRI study Donner et al. (2002) found that increased activity in parietal (and frontal) lobes in conjunction search could be attributed to increased difficulty, but some regions could not. Specifically, an area at the junction of the posterior inferior parietal sulcus and dorsal occipital lobe was active in tasks that required feature binding, independent of the search difficulty involved. Shafitz et al. (2002) also controlled for search difficulty to detect a target in conjunction and feature displays and linked spatial functions associated with parietal activation to binding errors and spatial processing.

In sum, there is converging evidence that the parietal lobes are involved in binding basic visual features that are encoded by specialized neural populations in the ventral stream of processing that have been associated with object formation (Ungerleider and Mishkin, 1982). Importantly, it is the spatial functions of the parietal lobe that seem to be involved. The data as a whole support predictions originally proposed by feature integration theory that spatial attention is important for accurate feature integration. They also demonstrate that feature binding in multi item displays does not take place normally without input from the parietal lobes.

0 0

Post a comment