The mnemonic value of perceptual identification.

In four experiments, subjects were required to name words presented on a CRT screen. On generate trials, the words were presented quickly, at a point where roughly half could be identified correctly; on read trials, the items were presented for a full second, allowing for rapid and easy naming. A surprise recognition test for the presented items then revealed a substantial retention advantage for the briefly presented items, but no similar advantage was produced in a recall. It is argued that under rapid viewing conditions subjects may fail to extract enough visual features to allow for immediate resolution, requiring the initiation of a kind of data-driven generation process. This latter process then produces a generation effect for the briefly presented items compared with the read items, but only on a retention test that shows sensitivity to data-driven processing. These results are discussed from the standpoint of current theoretical views on the generation effect.