Abstract
The capacity of visual information processing is limited in many aspects. Identification of the second visual target (T2) is impaired in rapid serial visual presentation (RSVP) when it is presented within half a second after the appearance of the first target (T1). This deficit in identifying T2, called attention blink (AB), is believed to reflect the capacity limitation in high-level processes such as attention and short-term memory consolidation.
Can training of RSVP tasks reduce or even eliminate capacity limitations? If so, as in perceptual learning, can this occur by mere repetitions of the task? In Exp 1, subjects were trained in a typical RSVP task by repeated performance: for 450 trials per day during a 3-consecutive-days period, subjects reported two digits (T1 and T2) presented with a fixed 200ms SOA (stimulus onset asynchrony) that were embedded in a sequence of alphabetic letters. Despite training, there was no evidence of AB reduction.
In Exp. 2, subjects performed a modified RSVP task identical to that of Exp 1 except that during training T2 was spotlighted red while T1 and all letters were white, thus attracting attention to T2. To assess training effects, AB was measured before and after each training session with an RSVP task in which all items including T2 were white. Surprisingly, AB was eliminated after a single day of training, and was continuously absent for several months.
These results indicate that even the deficit induced by capacity limitations, such as AB, can be overcome by training observers' attention to a blinked item, and that the learning effect lasts for a long time. Contrary to perceptual learning formed by mere exposure to some primitive visual features (Watanabe et al., 2001), reduction or elimination of capacity limitations in high-level processes may require focused attention.
This research was supported by NIH (R21 EY018925, R01 EY15980-04A2), NSF (BCS-0549036) and HFSP (RGP 18/2004.