A detailed description of the Simultaneous Type Serial Token (STST) model is presented. STST is a model of temporal attention and working memory, which encapsulates five principles: 1) Chun and Potter's (1995) 2-stage model; 2) a stage one Salience Filter; 3) Kanwisher'... View more
1. The Basic Blink. A typical AB serial-position curve, which arises from the letters-in-digits task, is shown in Figure 5. Points to note are, a. the blink is a 200ms – 500ms (approx) interval post T1 onset in which performance on T2, conditional on correct report of T1 (i.e. T2 | T1), is significantly below baseline; b. generally, the blink has a sharper onset than offset; and c. if T2 immediately follows T1 it is reported at baseline levels, which is described as lag 1 sparing. Of particular note is lag 1 sparing, which seems counter-intuitive, since one may believe that limited attentional resources are maximally engaged in processing T1 when T2 “arrives” at lag 1.
Insert Figure 5 Here
2. T1+1 and T1+2 blank. The blink was attenuated when a blank was placed in the T1+1 position, but not when it was placed at T1+2 (Chun & Potter, 1995; Raymond et al., 1992), see Figure 5.
3. T2 unmasked. The blink was significantly attenuated if T2 was the last item in the stream (Giesbrecht & Di Lollo, 1998), see Figure 5. In addition, we suggest that, placing blanks after T2 will attenuate the blink.
4. T1 Performance. T1 performance was reduced at lag 1; see Figure 6.
Insert Figure 6 Here
5. Swaps. It is possible that both T1 and T2 are identified, but they are perceived in the wrong order. In Chun and Potter (1995), such T1 - T2 swaps were maximal at lag 1 and effectively disappeared by lag 3; see Figure 6. In fact, at lag1, temporal order judgment was only a little above chance.