Some loss of location information at the level of the character buffer as a result of
masking is, nonetheless, envisaged in the dual-buffer model. Consequently, the mask-ISI
function in terms of p(L|I) is expected to rise with increases in mask-ISI, as may be seen
from the bottom left panel of Figure 5. The top left panel of Figure 5 reflects the expected
partial-report performance as a function of mask-delay in terms of the dual-buffer model.
The partial-report probe was always presented at the immediate offset of the array
in Experiment 3. The mask-ISI factor was applicable to the masking manipulation on half
the occasions. For this reason, the ISI will be called mask-ISI when Experiment 3 is
discussed. The five mask-ISI levels were 0, 150, 250, 500, and 750 ms. The no-masking
condition was analogous to Mewhort et al.'s (1981) Condition I when mask-ISI = 0 ms,
whereas the masking condition was analogous to their Condition 4. The stimulus duration
was 20 ms.
Results
Backward masking was not present in half of the session. For the purpose of data
analysis, the no-masking condition was treated as the control level of the mask-ISI
manipulation. That is, data were analyzed with a one-way ANOVA in which there were
six levels of mask-ISI.
In the two bottom panels of Figure 2 are depicted the subjects' intralist and
extralist intrusion errors. They should be compared to Mewhort et al.'s (1981) findings
under their Condition 4. By and large, the present observations are not inconsistent with
their findings in terms of intralist and extralist intrusion errors. In the case of both
extralist and intralist intrusions, the mask-ISI effect was significant; for intralist
intrusions, F(5, 65) = 6.05; for extralist intrusions, F(5, 65) = 4.98.
The subjects' partial-report performance in terms of correct recall of both item
identity and its intra-array location has been graphically depicted in the right top panel of
Figure 5. Performance increased with increases in mask-ISI under the masking condition.
The statistical analysis showed a significant mask-ISI effect, F(5, 65) = 12.27.
The availability of item information increased with increases in mask-ISI in the
presence of a mask (see the middle right panel of Figure 5). This is confirmed by a
significant mask-ISI effect, F(5, 65) = 19.26. The mask-ISI manipulation did not have
any effect on the availability of location information, as may be seen from the bottom
right panel of Figure 5.
Discussion
As in Experiments 1 and 2, the functional relation between (a) the subjects'
intralist intrusion errors and mask delay and (b) the subjects' extralist intrusion errors and
mask delay were very similar to those found by Mewhort et al. (1981).