Two experiments tested perceptual processing of Garner and Clement's (1963) good and poor five-dot patterns by the left and right hemispheres. Two (good) patterns were from four-member equivalence sets, and the other (poor) pattern was from an eight-member equivalence set. One of the good patterns formed a T shape that could be processed as a linguistic unit. In Experiment 1, 80 right-handed subjects made same-different judgments for lateralized 200-msec paired presentations of these patterns. When both presentations were to the RVF/LH, response latency was faster for the T pattern than for the other two. When both presentations were to the LVF/RH, response latency was faster for the two good patterns than for the poor pattern. When the first pattern was presented to the RVF/LH and the second was presented to the LVF/RH, response latencies were lower for the two good patterns than for the poor pattern. Also, when the first pattern was presented to the LVF/RH and the second was presented to the RVF/LH, response latency was faster for the T pattern than for the other good pattern which was, in turn, faster than for the poor pattern. Experiment 2 used a duration judgment task (Avant & Lyman, 1975) to test effects of pattern goodness on apparent durations of pre- and postmasked 10-msec pattern presentations. With left hemisphere inputs, presentations of good patterns were judged to be longer than presentations of the poor pattern. When each hemisphere compared the T and the other good pattern, presentations of the T pattern were judged to be longer, and the right hemisphere further discriminated among pattern orientations. Presentations of the T pattern to each hemisphere were judged to be longer than presentations of the poor pattern, and both hemispheres discriminated among orientations of both patterns. These results indicate that the two hemispheres can, during perceptual processing, function cooperatively, and that both prerecognition and conscious perceptual operations are guided by task demands.