Journal of Experimental Psychology:
Human Perception and Performance
2015, Vol. 41, No. 2, 269 –276
© 2015 American Psychological Association
0096-1523/15/$12.00 http://dx.doi.org/10.1037/xhp0000026
OBSERVATION
On the Independence of Visual Awareness and Metacognition:
A Signal Detection Theoretic Analysis
Barbara Jachs
Manuel J. Blanco
Imperial College London
Universidad de Santiago de Compostela
Sarah Grantham-Hill and David Soto
Imperial College London
Classically, visual awareness and metacognition are thought to be intimately linked, with our knowledge
of the correctness of perceptual choices (henceforth metacognition) being dependent on the level of
stimulus awareness. Here we used a signal detection theoretic approach involving a Gabor orientation
discrimination task in conjunction with trial-by-trial ratings of perceptual awareness and response
confidence in order to gauge estimates of type-1 (perceptual) orientation sensitivity and type-2 (metacognitive) sensitivity at different levels of stimulus awareness. Data from three experiments indicate that
while the level of stimulus awareness had a profound impact on type-1 perceptual sensitivity, the
awareness effect on type-2 metacognitive sensitivity was far lower by comparison. The present data pose
a challenge for signal detection theoretic models in which both type-1 (perceptual) and type-2 (metacognitive) processes are assumed to operate on the same input. More broadly, the findings challenge the
commonly held view that metacognition is tightly coupled to conscious states.
Keywords: metacognition, visual awareness, perceptual decision making
Persaud et al., 2011) and, in fact, metacognition has long been
strongly coupled with higher-level cognition, namely, working
memory, conscious awareness, and executive control processes
(Shimamura, 2008).
Accordingly, metacognitive ability ought to be dependent on factors that influence conscious awareness. For instance, the monitoring
of the correctness of our perceptual decisions ought to be better when
the perceptual saliency of the stimulus is stronger rather than weak.
Recent studies have however challenged the traditional view that
higher-level cognition, namely, reading, doing arithmetic, cognitive
control, and working memory, are by necessity contingent on conscious awareness (Dutta, Shah, Silvanto, & Soto, 2014; Hassin,
Bargh, Engell, & McCulloch, 2009; Pan, Lin, Zhao, & Soto, 2014;
Sklar et al., 2012; Soto, Mäntylä, & Silvanto, 2010; Soto & Silvanto,
2014; van Gaal, Ridderinkhof, Scholte, & Lamme, 2010).
Therefore here we sought to characterize more closely the
relationship between perception, conscious awareness, and metacognition of perceptual decision making. We assessed how type-1
(perceptual) and type-2 (metacognitive) processing are influenced
by levels of stimulus awareness, under conditions in which the
stimulus input is held constant but awareness fluctuates across
experimental trials.
To assess the dependence/independence of awareness and metacognition, we used a Gabor orientation discrimination task in
conjunction with different perceptual awareness scales (Overgaard, Timmermans, Sandberg, & Cleeremans, 2010) and response
confidence ratings (see Figure 1 and Experimental Procedures for
Experiment 1). The luminance of the Gabor was titrated in a
Metacognition refers to our ability to reflect upon our own
cognitive processes; for instance, to evaluate the quality of our
perceptions, thoughts and actions. Intuitively, metacognitive processes are thought to be a key aspect of conscious awareness, and
cognitive functions associated with reflection and introspection are
indeed difficult to conceive independently of consciousness.
Influential models of metacognition (Shimamura, 2008) highlight an interaction between first-order object-level processes (e.g.,
a task involving stimulus discrimination) and second-order monitoring processes that operate on the output of the first-order process; for instance, to provide a confidence estimate of the correctness of first-order perceptual choices. The contents of these
second-order metacognitive processes are often assumed to reflect
the contents of consciousness (Kunimoto, Miller, & Pashler, 2001;
This article was published Online First February 9, 2015.
Barbara Jachs, Department of Medicine, Division of Brain Sciences,
Imperial College London; Manuel J. Blanco, Facultad de Psicología, Laboratorio de Percepción y Rendimiento Humano, Universidad de Santiago de
Compostela; Sarah Grantham-Hill and David Soto, Department of Medicine, Division of Brain Sciences, Imperial College London.
We thank Steve Fleming for sharing the code to compute the area under
the type-2 ROC curve.
Correspondence concerning this article should be addressed to David
Soto, Imperial College London, Department of Medicine, Division of Brain
Sciences, Charing Cross Campus, St. Dunstan’s Road, W6 8RP, London,
UK. E-mail: [email protected]
269
JACHS, BLANCO, GRANTHAM-HILL, AND SOTO
270
Figure 1.
Example of an experimental trial: sequence of events.
precalibration stage using a 1-up 1-down staircase in order to
present observers with stimuli near the threshold of conscious
awareness. We derived receiver operating characteristics (ROC)
curves associated with both type-1 performance (i.e., sensitivity to
the Gabor orientation) and type-2 performance (i.e., sensitivity to
the correctness of the type-1 perceptual decisions). The paradigm
therefore allowed us to assess how awareness of the stimulus
influences both type-1 perceptual and type-2 metacognitive sensitivity.
Note that our key aim here was not to prove that metacognitive
operations can be successfully deployed nonconsciously (i.e.,
above chance type-2 sensitivity to response correctness for truly
subliminal items). We wondered instead whether type-2 sensitivity
(metacognition) is indeed strongly coupled to stimulus awareness;
were this the case, type-2 metacognitive sensitivity ought to covary
with stimulus awareness in a similar fashion as type-1 perceptual
sensitivity would be expected to covary with stimulus awareness.
Notably, according to signal detection theoretic models (Galvin,
Podd, Drga, & Whitmore, 2003) both type-1 and type-2 processes
are assumed to operate on the same input, which, in the present
experimental context, would reflect the output of neural channels
in early visual cortex tuned to Gabor orientation (Hubel & Wiesel,
1974; Thomas & Shimamura, 1975). From this, it follows that any
effect of awareness on performance should affect type-1 perceptual and type-2 metacognitive sensitivity to a similar extent.
Type-2 sensitivity is often suboptimal (i.e., lower than what would
be expected from the level of type-1 sensitivity), perhaps because
not all of the information available for the type-1 decision is used
for the type-2 metacognitive judgment (Maniscalco & Lau, 2012).
Even within this constraint, however, it would be expected that
awareness ought to influence type-1 and type-2 sensitivity to a
similar degree.
Experiment 1
Participants
The participants were 13 healthy volunteers (7 male, 6 female;
age range 22–30 years) who took part in return of £20. All
participants were naïve as to the experimental hypothesis, had
normal or corrected-to-normal vision, and provided written informed consent prior to the experiment. The study was approved
by the West London Research Ethics Committee.
Apparatus
The stimuli were presented on a gamma calibrated CRT display
(HP p1230, 1,600 ⫻ 1,200 pixels, 85 Hz refresh rate). The stim-
ulus display and responses were controlled using PsychoPy
(v.1.78.00) (Peirce, 2007), a psychophysics software written in
Python (http://www.psychopy.org/).
Experimental Procedures
The experiment took place in a darkened room. Displays were
viewed at a distance of approximately 90 cm. Each trial began with
a fixation display (a central ⫹ printed in black) for 500 ms
followed by a two-alternative forced choice (2AFC) discrimination
task for the orientation of a Gabor patch (3 cm diameter) presented
centrally on a gray background (luminance ⫽ 10.48 cd/m2). Figure
1 depicts the sequence of critical displays on a trial. Gabor orientation was 40° either to the left or right from the vertical; its spatial
frequency was 0.1 cycles per pixel. The Gabor was presented for
35 ms during the experimental trials and was immediately followed by a circular backward mask of black and white random
dots for 353 ms (11.34 cd/m2). Experimental trials were preceded
by different task training and stimulus calibration phases (see
below for further details).
Observers were first exposed to a training phase with masked
Gabors presented for 362 ms or for 35 ms, with a fixed luminance
of 11.97cd/m2 in order to familiarize participants with the orientation discrimination task. Feedback about the orientation discrimination response was given on a trial by trial basis.
Next, the luminance of the Gabor was calibrated using a 1-up:
1-down adaptive staircase. Observers were presented with a 35 ms
masked Gabor and asked to report whether they could see the
Gabor’s orientation or whether they were unaware of it. The
luminance of the Gabor was reduced following orientation-aware
trials and increased following orientation-unaware trials. Initial
Gabor luminance was set to 11.97cd/m2. The percentage of aware
responses was calculated on a trial by trial basis. Once the perceptual reports stabilized around the subjective awareness threshold (i.e., over 10 trials around 50% aware responses), we used the
contrast measurements of the Gabor items across the last 10 trials
to get an average, which was then used for the next procedural
stage.
Subsequently, observers received 30 training trials in the experimental task (see Figure 1) using 35 ms masked Gabors with the
luminance level established in the prior calibration phase. Participants were required to: (a) report their level of awareness of the
Gabor orientation (unaware, aware), (b) report the orientation of
the Gabor (2AFC), and (c) to rate the confidence about the correctness of their responses on a scale of 1–3. Participants were
instructed that confidence ratings should be conceived in a relative
manner; accordingly, observers were instructed that the full range
CONSCIOUS AWARENESS AND METACOGNITION
of these relative confidence estimates ought to be equally used
both on aware and unaware rated trials. Participants were encouraged to be as accurate as possible in their responses and reaction
time was not emphasized. Each of the three response frames was
presented during an unlimited time window until a response was
collected.
Prior to the experimental phase, the luminance of the Gabor was
recalibrated by employing the procedure described above with the
luminance output from the prior calibration as a starting point.
Lastly, participants performed 10 blocks of each 50 experimental trials, as depicted in Figure 1.
Signal Detection Theoretic (SDT) Analysis
All data processing and analyses were completed using R (R
Development Core Team, 2014) and Matlab. We used a receiveroperating-characteristic (ROC) analysis (Egan, 1975; Green &
Swets, 1966) to measure type-1 and type-2 sensitivity. Regarding
the type-1 perceptual sensitivity analyses, the “signal” and “noise”
were defined as left and right oriented Gabor, respectively. A “hit”
was therefore a correct response (“left”) to a left-oriented Gabor
and a “correct rejection” was a correct response (“right”) to a
right-oriented Gabor. A “false alarm” was an incorrect response
(left) to a right-oriented Gabor and a “miss” was an incorrect
response (right) to a left-oriented Gabor. To obtain the type-1 ROC
curves, we plotted the probability of hits as a function of the
probability of false alarms for all possible decision criteria. The
different points in the ROC curve were obtained by calculating
cumulative probabilities for hits and false alarms along a confidence continuum ranging from 3 (e.g., sure) to 1 (e.g., guess) in
signal-present trials (left-oriented Gabor), and from 1 (e.g., guess)
to 3 (e.g., sure) in noise trials (right-oriented Gabor). From simple
Figure 2.
271
geometry, we computed the area under the ROC curve (Kornbrot,
2006) as a distribution-free measure of the discriminability of the
two Gabors.
Regarding the type-2 sensitivity analyses, the area under the
ROC curve was calculated by plotting the cumulative probability
of correct responses (either hit or correct rejection) and the cumulative probability of incorrect responses (either false alarm or miss)
across the different levels of confidence. We estimated the area
under the ROC curve as a distribution-free measure of the discriminability of correct and incorrect type-1 performance or metacognitive ability (Fleming, Weil, Nagy, Dolan, & Rees, 2010). We
also computed type-1 d= and also meta-d= (Maniscalco & Lau,
2012), which is basically a parametric estimation of type-2 sensitivity predicted from the type-1 signal detection SDT model, done
by fitting a type-1 SDT model to the observed type-2 performance
data and estimating the type-2 ROC curves.
Results and Discussion
Figure 2 depicts the area under the type-1 and type-2 ROC
curves. Notably, performance in the “unaware” state was significantly above chance levels (0.5) both for type-1 and type-2 sensitivity, t(12) ⫽ 6.698, p ⬍ .0001 and t(12 ⫽ 3.205, p ⫽ .007,
respectively). Interestingly, while type-1 performance was better
in the aware relative to the unaware state, t(12) ⫽ 6.292, p ⬍
.0001, type-2 performance was not significantly different across
awareness states, t(12) ⫽ 0.545, p ⫽ .595.
The above observations were further confirmed by means of an
analysis of variance (ANOVA) with awareness (aware, unaware)
and performance type (perceptual, metacognitive) as factors. There
was a main effect of awareness F(1, 12) ⫽ 13.6 p ⫽ .0031)
showing that overall performance was, overall, better in the aware
Type-1 and type-2 ROC curves as a function of the level of awareness.
JACHS, BLANCO, GRANTHAM-HILL, AND SOTO
272
state. There was also a main effect of performance type, showing
that type-1 sensitivity outperforms type-2 sensitivity F(1, 12) ⫽
88.87, p ⬍ .00001). Critically, there was a significant interaction
between factors F(1, 12) ⫽ 24.39, p ⬍ .0003, in keeping with the
observation that awareness effects were higher on type-1 relative
to type-2 sensitivity. Type-2 sensitivity was similar in aware and
unaware states, t(12) ⫽ 0.545, p ⫽ .595. Identical results were
found with type-1 d= and meta-d= (Maniscalco & Lau, 2012):
awareness state interacted with performance type, F(1, 12) ⫽
11.17, p ⬍ .005, so that awareness effects were far higher on
type-1 d= than on meta-d=; meta-d= was not different across awareness conditions, t(12) ⫽ 0.82, p ⫽ .428 (Figure 3).
These results suggest that metacognition of perceptual decisionmaking, assessed by type-2 sensitivity scores from SDT, is not
strongly dependent on the level of stimulus awareness. Notably,
object perception, assessed by type-1 sensitivity was strongly
dependent on stimulus awareness. Before delving into further
discussion, we wished to replicate this pattern of results by using
a more exhaustive measure of stimulus awareness. Experiment 2
was devised with this in mind.
Experiment 2
Participants
Eighteen new healthy volunteers (9 male; aged 20 –29 years)
took part. One participant was excluded due to overall chance
performance in the orientation discrimination task. Another participant was excluded because he had only 24 2-rating awareness
trials, hence precluding accurate calculation n of type-2 sensitivity.
One participant was excluded for not using the full range of the
confidence scale (i.e., no 3-confidence ratings reported). However,
the pattern of results remained unchanged when all data were
analyzed.
Experimental Procedures
The experimental procedure was similar to Experiment 1, except
that here we used a perceptual awareness scale following Sandberg, Bibby, Timmermans, Cleeremans, & Overgaard, 2011 (see
their Experiment 2), which provides a more exhaustive measure of
awareness than the binary scale used in Experiment 1. Awareness
was rated by observers on a trial by trial basis, such that 1-rating
corresponded to “no experience of the stimulus,” 2-rating corresponded to “a brief glimpse but no experience of the orientation,”
and 3-rating signified “almost clear experience or clear experience
of the Gabor orientation.”
Results
ROC analyses of type-1 and type-2 sensitivity were not performed on trials with reports of almost clear or vivid visual
experience because these were a minority of the total and hence
precluded accurate derivation of the area under the ROC curve. In
keeping with Experiment 1, performance in the “unaware” state
(e.g., trials with ratings of “No experience of the stimulus”) was
significantly above chance levels (0.5) both for type-1 and type-2
sensitivity, t(14) ⫽ 5.15, p ⬍ .0001 and t(14 ⫽ 4.3, p ⬍ .0001,
respectively). Figure 4 illustrates type-1 and type-2 ROC curves.
An ANOVA of the areas under the ROC curves with awareness
and performance type as factors showed the main effects of awareness, F(1, 14) ⫽ 15.2, p ⫽ .0016 and performance type, F(1, 14) ⫽
35.81, p ⬍ .00001. More critically, replicating the findings from
Experiment 1, there was a significant interaction between factors
F(1, 14) ⫽ 24.85, p ⬍ .0001), indicating that the difference in
sensitivity between aware and unaware states was stronger for
type-1 relative to type-2 performance. We also note that type-2
sensitivity did not differ across awareness states, t(14) ⫽ 1.549,
p ⫽ .143. The same interaction effect between awareness and
performance type was found on meta-d=, F(1, 14) ⫽ 23.47, p ⬍
.0002 (Figure 5). Here, meta-d= results showed a significant awareness effect on type-2 sensitivity, t(14) ⫽ 2.834, p ⫽ .013. However, we note this result was absent in Experiment 1 and also in
Experiment 3 below, and, in any case, it does not challenge our
main conclusion that awareness effects are strongest on type-1
sensitivity and greatly reduced on type-2 sensitivity.
Experiment 3
Figure 3. Type-1 d= and type-2 meta-d= scores as a function of the level
of awareness.
ROC analyses provide a measure of sensitivity which is biasfree from a SDT perspective; hence, type-2 sensitivity ought to be
independent on a particular tendency of the individual to report
(e.g.) high confidence. The same argument applies to type-1 d= and
meta-d= (Maniscalco & Lau, 2012). However, it remains to be seen
in the present study whether visual awareness could have had any
effect on decision biases beyond the modulation of type-1 sensi-
CONSCIOUS AWARENESS AND METACOGNITION
Figure 4.
273
Type-1 and type-2 ROC curves as a function of the level of awareness.
tivity. The present Experiment 3 tested for awareness effects on
type-2 bias (e.g., the observer’s tendency to report high confidence). We note that the design of Experiments 1 and 2 did not
allow us to derive a consistent SDT-based measurement of type-2
bias because there was an uneven number of three confidence
ratings and the computation of type-2 bias can be biased with an
odd number of ratings (Kornbrot, 2006). To provide an unbiased
measure of type-2 bias we here used an even number of 2 ratings
along a continuum of relative confidence.
Participants
Ten new healthy volunteers (5 male; aged 21–25 years) took
part.
Experimental Procedures
The experimental procedure was similar to the previous experiments. The main difference is that the awareness and confidence
scales were simplified to facilitate the computation of type-2
confidence biases. In reporting confidence, participants were asked
to inform how confident there were about the correctness of the
orientation response on a relative scale of 1 (relatively less)–2
(relatively more) confident. In reporting awareness, participants
were asked to report “unawareness” when they had no experience
of the Gabor or saw a brief glimpse but were not aware of the
orientation, and to report “awareness” when they could see the
Gabor’s orientation somewhat or almost clearly. There were eight
blocks of 50 trials each.
Signal Detection Theoretic (SDT) Analysis
We used nonparametric assessments of type-2 sensitivity and
type-2 bias because the parametric assumptions may not hold for
type-2 data (Galvin et al., 2003). We used A= (Macmillan &
Creelman, 1991) to derive a measure of both type-1 and type-2
sensitivities. For the type-1 sensitivity analyses, the hits and false
alarms were defined as in the previous experiments. For the type-2
sensitivity analyses, the signal was the correct response and the
noise the incorrect response. Accordingly, a type-2 hit was defined
as a correct response (either a type-1 hit or a type-1 correct
rejection) with high confidence and a type-2 false alarm was
defined as an incorrect response (either a type-1 false alarm or a
miss) with high confidence. Type-2 correct rejections were incorrect trials with low confidence and type-2 misses were correct
trials with low confidence. We also report d=/meta-d= results
(Maniscalco & Lau, 2012).
To estimate the type-2 bias, we used B=k (Kornbrot, 2006), and
also B”D (Donaldson, 1992) which has been shown to outperform
other nonparametric estimates of bias in certain experimental settings (e.g., vigilance) (See, Warm, Dember, & Howe, 1997).
Results and Discussion
Analyses of type-1 and type-2 sensitivity confirmed the results
found in the previous two experiments. Type-1 A= was higher on
aware relative to unaware trials, t(9) ⫽ 4.126, p ⫽ .0025, which
did not differ in terms of type-2 sensitivity, t(9) ⫽ 0.998, p ⫽ .344.
Similar effects were found on d= and meta-d=. Type-1 d= was
enhanced on aware relative to unaware trials, t(9) ⫽ 3.159, p ⫽
.011 but this was not the case for meta-d=, t(9) ⫽ 0.224, p ⫽ .827.
Critically, this pattern of results was found in the absence of
awareness effects on type-2 bias, t(9) ⫽ 0.025, p ⫽ .98, for B=k
and t(9 ⫽ 0.2, p ⫽ .84, for B==D). These results are depicted in
Figures 6 and 7.
These data indicate that awareness effects occur in the absence
of effects on type-2 decision biases.
274
JACHS, BLANCO, GRANTHAM-HILL, AND SOTO
Figure 5. Type-1 d= and type-2 meta-d= scores as a function of the level
of awareness.
General Discussion
We found that visual awareness had a profound effect on type-1
perceptual sensitivity but, intriguingly, the awareness effect on
type-2 metacognitive sensitivity was far lower by comparison.
Notably, metacognitive sensitivity was significantly above chance
across awareness states. The disproportionate effect of awareness
on type-1 relative to type-2 sensitivity was observed across three
experiments, which varied the observer’s criteria to report the level
of perceptual awareness. This pattern of results is newsworthy in
several theoretical junctures.
Traditionally, the information that is appraised in the metacognitive process is typically thought to reflect the contents of conscious visual experience (Kunimoto et al., 2001; Persaud, McLeod,
& Cowey, 2007) and intuitively it is difficult to think how an
observer can evaluate his or her responses to items independently
of his or her level of awareness of the items in question. The
present data suggests that the contents of metacognitive computations are dissociable from levels of stimulus awareness. Influential
models (Shimamura, 2008) conceive metacognition as resulting
from the interaction between perceptual processes operating on
sensory representations (e.g., a feature discrimination task) and
second-order (evaluative) processes that monitor the output of the
object-level process (e.g., evaluation of the correctness of type-1
decisions). The present findings challenge this notion by showing
that awareness can have distinctive effects on type-1 and type-2
processes.
“Blindsight” patients report no visual experience for items presented in the visual field impaired by their lesion in primary visual
cortex lesions but nevertheless can display astonishing high levels
of perceptual sensitivity. Blindsight has provided remarkable insight into the relation between perception and awareness by showing that awareness may not be necessary for type-1 perceptual
discrimination functions. Interestingly, there is some data suggesting that blindsight patients can sometimes report higher levels of
decision confidence (e.g., by placing a higher wager) following
correct than incorrect perceptual decision for targets presented in
the “blind” visual field (Persaud et al., 2011; Persaud et al., 2007),
which suggests that confidence ratings may be diagnostic of discrimination accuracy to some degree, even in the absence of
awareness.
To the best of our knowledge, only a recent study in healthy
human observers assessed type-1 and type-2 sensitivity as a function of stimulus awareness (Charles, Van Opstal, Marti, & De-
Figure 6. A= scores for the type-1 and type-2 task (left panel) and d=/meta-d= scores (right panel) as a function
of the level of awareness.
CONSCIOUS AWARENESS AND METACOGNITION
275
Figure 7. Type-2 decision bias scores: (B=k, left panel; B==D, right panel). Negative values of B= k reflect a
relative bias toward lower confidence ratings while positive values correspond to higher confidence biases. B==D
values equal to 0 indicates no bias; positive numbers represent a tendency report little confidence and negative
numbers represent a tendency to report high confidence.
haene, 2013). Charles and colleagues showed that type-2 metacognitive evaluation for nonconscious targets in a number comparison
task could exceed chance levels. Unlike here, however, Charles et
al. (2013) also found that type-2 metacognitive sensitivity was
significantly higher than type-1 perceptual sensitivity. Typically,
as we found here, type-2 sensitivity tends to be low (e.g., suboptimal) relative to type-1 sensitivity—that is, not all the information
available for the type-1 process may be available for the secondorder metacognitive judgment (Maniscalco & Lau, 2012). Charles
et al. (2013) rightly argued that higher type-2 sensitivity in their
experimental protocol could reflect additional accumulation of
information following the type-1 decision (Resulaj, Kiani, Wolpert, & Shadlen, 2009), which would continue to inform, therefore,
the type-2 process. This would lead to errors or hits being greater
acknowledged following the type-1 decision and hence leading to
a boost in type-2 sensitivity.
Our findings are in keeping with the Charles et al. (2013) study
in that stimulus awareness does not appear to be a prerequisite for
type-2 metacognitive sensitivity. Importantly, the present findings
go beyond to demonstrate that type-1 and type-2 processes can
dissociate in how they relate to stimulus awareness. The attenuated
effect of awareness on type-2 sensitivity in the presence of a
significant awareness effect on type-1 sensitivity suggests that
type-2 metacognitive operations are not necessarily carried out on
the same type of representation or follow a similar process to that
underlying type-1 performance. These results, therefore, place
critical constraints on extant signal detection theoretic frameworks,
which postulate that type-1 and type-2 sensitivity share a common
informational source which determines the sensitivity of type-2
decisions (Clifford, Arabzadeh, & Harris, 2008; Galvin et al.,
2003; Maniscalco & Lau, 2012; Pleskac & Busemeyer, 2010).
One limitation of the present study is that behavior was modeled
by using a static analytical model of the decision process (i.e.,
SDT). Because SDT cannot capture the dynamics of the decision
process, it may well be that the relationship between metacognition
and awareness obey more complex rules than the degree of dependency or independency. To tackle this issue, dynamic models
of the decision process such as drift diffusion or accumulator
models (Brown & Heathcote, 2008; Ratcliff & McKoon, 2008)
may be useful to better understand how awareness relates to type-1
and type-2 processes; however, any attempt to assess the role of
awareness in this regard will be confronted with the difficulty of
separating type-1 and type-2 processes because these are likely to
overlap in time. Novel paradigms need to be developed in order to
tackle this issue.
The present findings challenge existing accounts of metacognition and its relation to perception and conscious awareness by
suggesting that stimulus awareness and first-order perceptual processes are not necessarily key determinants in the computation of
metacognitive judgments. These findings are consonant with the
view that higher-order processes associated to working memory
and cognitive control may be dissociated to some extent from
conscious awareness.
References
Brown, S. D., & Heathcote, A. (2008). The simplest complete model of
choice response time: Linear ballistic accumulation. Cognitive Psychology, 57, 153–178. http://dx.doi.org/10.1016/j.cogpsych.2007.12.002
Charles, L., Van Opstal, F., Marti, S., & Dehaene, S. (2013). Distinct brain
mechanisms for conscious versus subliminal error detection. NeuroImage, 73, 80 –94. http://dx.doi.org/10.1016/j.neuroimage.2013.01.054
Clifford, C. W. G., Arabzadeh, E., & Harris, J. A. (2008). Getting technical
about awareness. Trends in Cognitive Sciences, 12, 54 –58. http://dx.doi
.org/10.1016/j.tics.2007.11.009
Donaldson, W. (1992). Measuring recognition memory. Journal of Experimental Psychology: General, 121, 275–277. http://dx.doi.org/10.1037/
0096-3445.121.3.275
Dutta, A., Shah, K., Silvanto, J., & Soto, D. (2014). Neural basis of
non-conscious visual working memory. NeuroImage, 91, 336 –343.
http://dx.doi.org/10.1016/j.neuroimage.2014.01.016
Egan, J. P. (1975). Signal detection theory and ROC-analysis. New York,
NY: Academic Press.
Fleming, S. M., Weil, R. S., Nagy, Z., Dolan, R. J., & Rees, G. (2010).
Relating introspective accuracy to individual differences in brain structure. Science, 329, 1541–1543. http://dx.doi.org/10.1126/science
.1191883
276
JACHS, BLANCO, GRANTHAM-HILL, AND SOTO
Galvin, S. J., Podd, J. V., Drga, V., & Whitmore, J. (2003). Type 2 tasks
in the theory of signal detectability: Discrimination between correct and
incorrect decisions. Psychonomic Bulletin & Review, 10, 843– 876.
http://dx.doi.org/10.3758/BF03196546
Green, D., & Swets, J. (1966). Signal detection theory and psychophysics.
Huntington, NY: Wiley.
Hassin, R. R., Bargh, J. A., Engell, A. D., & McCulloch, K. C. (2009).
Implicit working memory. Consciousness and Cognition: An International Journal, 18, 665– 678. http://dx.doi.org/10.1016/j.concog.2009.04
.003
Hubel, D. H., & Wiesel, T. N. (1974). Sequence regularity and geometry
of orientation columns in the monkey striate cortex. The Journal of
Comparative Neurology, 158, 267–293. http://dx.doi.org/10.1002/cne
.901580304
Kornbrot, D. E. (2006). Signal detection theory, the approach of choice:
Model-based and distribution-free measures and evaluation. Perception
& Psychophysics, 68, 393– 414. http://dx.doi.org/10.3758/BF03193685
Kunimoto, C., Miller, J., & Pashler, H. (2001). Confidence and accuracy of
near-threshold discrimination responses. Consciousness and Cognition:
An International Journal, 10, 294 –340. http://dx.doi.org/10.1006/ccog
.2000.0494
Macmillan, N. A., & Creelman, C. D. (1991). Detection Theory: A User’s
guide. New York: Cambridge University Press.
Maniscalco, B., & Lau, H. (2012). A signal detection theoretic approach
for estimating metacognitive sensitivity from confidence ratings. Consciousness and Cognition: An International Journal, 21, 422– 430.
http://dx.doi.org/10.1016/j.concog.2011.09.021
Overgaard, M., Timmermans, B., Sandberg, K., & Cleeremans, A. (2010).
Optimizing subjective measures of consciousness. Consciousness and
Cognition: An International Journal, 19, 682– 684. http://dx.doi.org/
10.1016/j.concog.2009.12.018
Pan, Y., Lin, B., Zhao, Y., Soto, D. (2014). Working memory biasing of
visual perception without awareness. Attention Perception & Psychophysics, 76, 2051–2062. http://dx.doi.org/10.3758/s13414-013-0566-2
Peirce, J. W. (2007). PsychoPy—Psychophysics software in Python. Journal of Neuroscience Methods, 162, 8 –13. http://dx.doi.org/10.1016/j
.jneumeth.2006.11.017
Persaud, N., Davidson, M., Maniscalco, B., Mobbs, D., Passingham, R. E.,
Cowey, A., & Lau, H. (2011). Awareness-related activity in prefrontal
and parietal cortices in blindsight reflects more than superior visual
performance. NeuroImage, 58, 605– 611. http://dx.doi.org/10.1016/j
.neuroimage.2011.06.081
Persaud, N., McLeod, P., & Cowey, A. (2007). Post-decision wagering
objectively measures awareness. Nature Neuroscience, 10, 257–261.
http://dx.doi.org/10.1038/nn1840
Pleskac, T. J., & Busemeyer, J. R. (2010). Two-stage dynamic signal
detection: A theory of choice, decision time, and confidence. Psychological Review, 117, 864 –901. http://dx.doi.org/10.1037/a0019737
Ratcliff, R., & McKoon, G. (2008). The diffusion decision model: Theory
and data for two-choice decision tasks. Neural Computation, 20, 873–
922. http://dx.doi.org/10.1162/neco.2008.12-06-420
R Development Core Team. (2014). The R Project: A language and
environment for statistical computing. Vienna, Austria: R Foundation;
http://www.r-project.org
Resulaj, A., Kiani, R., Wolpert, D. M., & Shadlen, M. N. (2009). Changes
of mind in decision-making. Nature, 461, 263–266. http://dx.doi.org/
10.1038/nature08275
Sandberg, K., Bibby, B. M., Timmermans, B., Cleeremans, A., & Overgaard, M. (2011). Measuring consciousness: Task accuracy and awareness as sigmoid functions of stimulus duration. Consciousness and
Cognition: An International Journal, 20, 1659 –1675. http://dx.doi.org/
10.1016/j.concog.2011.09.002
See, J. E., Warm, J. S., Dember, W. N., & Howe, S. R. (1997). Vigilance
and Signal Detection Theory: An empirical evaluation of five measures
of response bias. Human Factors, 39, 14 –29. http://dx.doi.org/10.1518/
001872097778940704
Shimamura, A. P. (2008). A neurocognitive approach to metacognitive
monitoring and control. In J. Dunlosky & R. A. Bjork (Eds.), Handbook
of metamemory and memory (pp. 373–390). New York, NY: Psychology
Press.
Sklar, A. Y., Levy, N., Goldstein, A., Mandel, R., Maril, A., & Hassin,
R. R. (2012). Reading and doing arithmetic nonconsciously. PNAS
Proceedings of the National Academy of Sciences of the United States
of America, 109, 19614 –19619. http://dx.doi.org/10.1073/pnas
.1211645109
Soto, D., Mäntylä, T., & Silvanto, J. (2011). Working memory without
consciousness. Current Biology, 21(22), R912–R913. http://dx.doi.org/
10.1016/j.cub.2011.09.049
Soto, D., & Silvanto, J. (2014). Reappraising the relationship between
working memory and conscious awareness. Trends in Cognitive Sciences, 18, 520 –525. http://dx.doi.org/10.1016/j.tics.2014.06.005
Thomas, J. P., & Shimamura, K. K. (1975). Inhibitory interaction between
visual pathways tuned to different orientations. Vision Research, 15,
1373–1380. http://dx.doi.org/10.1016/0042-6989(75)90193-5
van Gaal, S., Ridderinkhof, K. R., Scholte, H. S., & Lamme, V. A. (2010).
Unconscious activation of the prefrontal no-go network. The Journal of
Neuroscience, 30, 4143– 4150. http://dx.doi.org/10.1523/JNEUROSCI
.2992-09.2010
Received August 26, 2014
Revision received October 27, 2014
Accepted November 24, 2014 䡲