Models of substitution masking

1
ECVP 2011, August 28 September 1, Toulouse,
France
Models of substitution masking Endel
PõderInstitute of Psychology, University of
Tartu, Estonia E-mail endel.poder_at_ut.ee
Background Substitution masking Enns and Di Lollo
(1997) found that a visual target can be strongly
masked by just four dots presented after the
target. A necessary precondition was unfocused
attention. Di Lollo, Enns Rensink (2000)
demonstrated a similar type of masking when the
masking dots were turned on simultaneously with
the target but remained visible for some time
after the target termination (common-onset
masking). Unfocused attention was warranted by
distracting objects. Computational Model of
Object Substitution (CMOS) Di Lollo et al (2000)
proposed that substitution masking reflects a
reentrant hypotheses-testing process in vision.
They presented a quantitative model (CMOS) that
should simulate this reentrant hypotheses-testing.
The new model Similar to CMOS, but with a more
realistic mechanism of attention. Two
attentional episodes in each trial. At the
beginning, attention is divided between all
objects in a display. The time of focusing of
attention is independent of set-size.

1. Unselective stage d dependent on set-size,
   independent on mask duration

1. Selective stage d dependent on mask duration
   (function estimated from the data), independent
   of set-size.

Total d
This study I evaluate the assumptions of the Di
Lollo et al (2000) model and its supposed
relationship with reentrant processing, and test
an alternative model for substitution masking
data.
Probability correct
Fit to the experimental data from Di Lollo et al
(2000) R20.99 (cf. CMOS, R20.93)
1. CMOS as an attentional gating model
There are two main equations for transformation
of signals from target, masker and noise
(1)
(2)
Substituting Wj(k-1) in equation (1) with
equation (2) transforms this part of the model
into a single equation
This is just an integrating circuit (low-pass
filter) applied to input signals!
Additional component delay of focusing of
attention, proportional to set-size
Probability of correct recognition is determined
by the target signal energy relative to the total
signal energy at the moment of arrival of
attention.
The low-pass input filter together with the
attentional delay form a simple attentional
gating model (Sperling Weichselgartner, 1995).
Conclusion 1 CMOS is a version of attentional
gating model. It does not contain anything
related to reentrant hypotheses-testing.
Estimated d of selective stage
2. New model for substitution masking
A problem with CMOS CMOS predicts a particular
form of the mask duration and set-size
interaction. Because the time of deployment of
attention is supposed to be proportional to the
set-size, the breakpoints of the masking curves
should be shifted along the x-axis, in proportion
to the set-size. This is not observed in
experimental data. Furthermore, there is no good
reason for the delay proportional to the
set-size.
Typical CMOS predictions
Conclusion 2 The new attentional gating model
fits the data better than CMOS.
References Di Lollo, V., Enns, J.T., Rensink,
R.A. (2000). Competition for consciousness among
visual events The psychophysics of reentrant
visual processes. Journal of Experimental
Psychology General, 129 (4), 481-507. Enns,
J.T., Di Lollo, V. (1997). Object substitution
A new form of visual masking in unattended visual
locations. Psychological Science, 8,
135-139. Sperling, G., Weichselgartner, E.
(1995). Episodic theory of the dynamics of
spatial attention. Psychological Review, 102,
503-532.