Shuyao Wang

We investigated how visual cognitive processes unfold under temporal
constraints using the rapid serial visual presentation (RSVP) paradigm. In a typical
RSVP task, stimuli are presented sequentially at the same spatial location, usually
at a rate of 10 items per second. This paradigm imposes high temporal pressure on
the visual system while effectively controlling other variables, such as spatial
confounds. Therefore, RSVP is a suitable tool to investigate how cognitive
resources are allocated to process items in quick succession.

An interesting phenomenon that emerges in the dual-target RSVP
paradigm is the attentional blink (AB). The AB refers to the impairment in
reporting a second target (T2) when it appears shortly after the first target (T1),
usually between 200 and 500 ms. The AB provides an ideal platform to examine
processing limitations because it allow for a direct comparison between two time
interval conditions: 1) short intervals, during which the identification of T2 is
significantly impaired; 2) longer intervals, during which no ‘blink’ occurs and T2 is
reported normally.

This dissertation hence utilizes the RSVP and AB paradigms to examine the
visual cognition under temporal constraints with regard to three specific aspects:
1. Whether working memory consolidation failure during the AB is a
discrete, all-or-none process, or a gradual loss of representational quality (Chapter
2).
2. If spatial attention can be flexibly adjusted under time pressure and how
this adjustment influences the discrete or graded nature of conscious awareness
(Chapter 3).
3. How internal memory representations guide attention and whether this
process can be captured by physiological signals for covert information detection
(Chapter 4).

In Chapter 2, we analyzed data derived from continuous report tasks
embedded in the AB paradigm to investigate the nature of working memory failures.
The analysis included four distinct datasets: three obtained from different
laboratories (Asplund et al., 2014; Karabay et al., 2022; Tang et al., 2020) and one
from a new experiment conducted for this study. We conducted a systematic
comparison of seven computations models, including “slot” models (standard
mixture, slot, slots plus resource models) and variable precision models. Our
results consistently showed that models incorporating a discrete guessing
component (“slot” models) provided the best fit at both group and individual levels.
Simulation analyses confirmed that these models accurately reproduce the “blink”
pattern. In contrast, the standard VP model failed to capture impaired
representations during the blink window, unless a guessing parameter was added.
These findings support the view that working memory (WM) consolidation in the
tested tasks is a capacity-limited process where targets either enter memory
successfully or fail entirely, with failures resulting in pure guessing.

In Chapter 3, we investigated whether participants can adaptively control
their attentional scale under temporal constraints. In this study, we assumed that
changes in the nature of T2 awareness, specifically indexed by its representational
precision, would indicate whether attentional scaling had been successfully
implemented. We manipulated the T1 location using session-wise, trial-wise, or
statistical regularities. We then analyzed T2 performance using mixture modeling
to dissociate the probability of conscious access from representational precision.
Our results showed that precision decreased when session-wise cues and statistical
regularities were implemented, indicating that a broad attentional scale was
successfully induced in these contexts. In contrast, trial-wise cues failed to
modulate precision, suggesting that immediate, trial-by-trial adaptation is not
possible in this condition. These findings demonstrate that while spatial attention
can be adjusted to influence the nature of conscious awareness, such spatial control
depends on stable explicit cues or implicit learning of regularities rather than
immediate adjustments.

In Chapter 4, we applied the RSVP paradigm to the practical domain of
concealed information detection to investigate how internal memory
representations influence temporal attention. Specifically, we embedded personally
familiar faces as task-irrelevant probes within the RSVP streams within a gender-
based target detection task. We then utilized four physiological measures: ERP (P3)
amplitude, theta-band power, pupil size, and pupil size change to assess covert
recognition. We found that all indicators successfully distinguished familiar probes
from neutral control stimuli at the group level. However, at the individual level,
although ERP and theta-band power signals provided reliable detection for most
participants, pupillary responses were less effective. This is likely due to the
difficulty of isolating relatively slow-developing pupillary signals in high-speed
RSVP paradigms. Furthermore, by integrating neural and pupillary signals, we
found that combined inference compensated to some extent for limitations of
single measures.

In sum, these chapters explore how the human cognitive system processes
information under temporal constraints using computational modeling, behavioral
performance, and physiological measures. Across three empirical studies, we
demonstrate that the cognitive system regulates information flow via reactive
gating, proactive spatial scaling, and memory-driven priority signals. Specifically,
our findings show that while working memory and conscious access in typical AB
tasks are primarily governed by discrete, all-or-none capacity limits, the nature of
awareness remains flexible and can be proactively configured by adjusting the
attentional scale. Furthermore, internal memory representations can automatically
capture processing resources even without overt report, a process that can be
reliably tracked through neural signals. Overall, this work suggests that temporal
constraints provide a window into the dynamic nature of human cognition,
illustrating how this system adaptively manages information processing under such
pressure.