The influence of psychophysical technique on the emergence of visual awareness

The Influence of Psychophysical Technique on

the Emergence of Visual Awareness

Rianne Lam (10772065)

Supervisor: Timo Stein Course: Bachelor thesis Date: 06/2017

Contents

1

Abstract

In the present study, the influence of psychophysical technique on the emergence of visual awareness was investigated. There is an ongoing debate whether the

emergence of consciousness can be best described as gradual or all-or-none. Recent literature points out that the psychophysical technique used to establish subjective awareness ratings could have an influence on the distribution of these ratings, and thus the emergence of visual awareness. In this study participants performed a localization task and rated target visibility on a subjective awareness scale, while either the attentional blink, backward masking or breaking-continuous flash suppression was used as a psychophysical technique. When looking at the results, the distribution of subjective awareness ratings was skewed to the lower visibility ratings for all psychophysical techniques, yet all possible subjective awareness

ratings were selected. This finding points out that the emergence of visual awareness is most likely to be more gradual than all-or-none, and that psychophysical technique does not seem to have an influence on this phenomenon.

2

Introduction

For a considerable time, consciousness has been a favored and appealing topic of research in many fields of psychology. A probable explanation for this is that consciousness is a vague and complex phenomenon that cannot be readily defined, yet it is vital for every human being. For example, consciousness can be described as “the quality or state of being aware especially of something within oneself1_{.” Despite} the definition of consciousness being uncertain, consciousness still is a concept that is central to every aspect of human behavior and thought. Therefore, many questions in respect to cognitive psychology can be explained to a certain extent by studying consciousness. However, it seems reasonable that before doing so it is important to first try to further clarify the concept of consciousness itself. An example of such need for clarification concerns the emergence of consciousness.

There is an ongoing debate whether the emergence of consciousness can be best described as a gradual or all-or-none phenomenon. On the one hand, researchers plead the emergence of consciousness to be a discrete process, where a stimulus fully enters awareness after exceeding some sort of threshold (Asplund, Fougnie, Zughni, Martin & Marois, 2014; Vul, Hanus & Kanwisher, 2009; Sergent & Dehaene, 2004b). Consciousness is simply there, or it is not. This idea originates from the global

neuronal workspace theory (Sergent & Dehaene, 2004a), which states that a stimulus must reach a certain threshold to activate brain areas so that it can enter awareness. On the other hand, there are also researchers who argue that the emergence of consciousness is a gradual process, where a stimulus enters awareness in a graded fashion (Bar, 2011; Overgaard, Rote, Mouridsen & Ramsoy, 2006; Nieuwenhuis & de Kleijn, 2011). By perceiving the emergence of consciousness as a gradual process, “in-between” states of awareness would seem to exist.

To establish whether consciousness can be marked as all-or-none or gradual, it is important to formulate a measure which optimally reflects the subjective

awareness of the participant in multiple settings. Likewise, there is some debate on which measure is best to be used to study the emergence of consciousness. There are some well-known measures to determine subjective awareness that will shortly be reviewed.

Firstly, subjective awareness can be measured using a continuous scale. In this measure, the participant is asked to select a point on a continuous scale that matches best with the subjective visibility of a potentially shown target. Only the two extreme

3 ends of the scale are labeled (e.g. “not visible” and “visible”). If participants would only choose the extreme ends of the scale this would indicate visual awareness to be all-or-none. If participants would also choose the intermediate points of the scale this would indicate visual awareness to be gradual, because this would mean that

participants sometimes did not see the target, but did perceive some of its features and thus the target had gained some, but not full consciousness. In research by

Sergent and Dehaene (2004b) and Del Cul, Baillet and Dehaene (2007) this measure of subjective awareness was used and they found a bimodal rating pattern, hence

concluding consciousness to be all-or-none.

Nevertheless, critiques on this measure concern the accessibility of the intermediate points on a continuous scale (Overgaard, Rote, Mouridsen & Ramsoy, 2006; Nieuwenhuis & de Kleijn, 2011). It could be that participants are inclined to only select the extreme ends of the scale because the center section of the scale doesn’t have sufficient meaning to them. On a continuous scale the intermediate points are completely open to interpretation. This would encourage participants to choose the extreme ends of the scale when maybe they actually were unsure about the visibility of the target.

Secondly, subjective awareness can be measured using a discontinuous scale. An often-used scale is the Perceptual Awareness Scale (Ramsoy & Overgaard, 2004). This scale consists of four labeled points (from “no experience” to “a clear

experience”). Another widely used scale are confidence ratings (Sandberg,

Timmermans, Overgaard & Cleeremans, 2010). Confidence ratings closely resemble the Perceptual Awareness Scale: principally they consist of a few labeled points that reflect the level of confidence in having perceived something. These scales differ from continuous scales in the way that every point on the scale has its label and the participant has less options to choose from in general. This difference obviates the above-mentioned critique about the options not being evident enough. Just as with the continuous scale, when participants only select the extreme ends of the scale this would suggest visual awareness to be all-or-none. In the other case, when

participants also select the intermediate points of the scale this would suggest visual awareness to be gradual. Overgaard, Rote, Mouridsen and Ramsoy (2006) performed a study using the Perceptual Awareness Scale and found a uniform rating pattern, thus concluding consciousness to be gradual.

However, a critique on this measure specifically concerns the scale consisting of just a few points to choose from, which could lead to a scale bias (Windey,

4 use all the scale points to fulfill the assumed demands of the experiment. If this is the case, subjective awareness would falsely be considered gradual.

Thirdly, subjective awareness can be measured using post-decision wagering (Persaud, McLeod & Cowey, 2007). In this measure participants are asked if a

potentially shown target was absent or present, and if they would bet a certain amount of money on it. By using this measure participants are generally more encouraged to make use of every possible option that is given, given that there is money at stake. An example is that a participant would think the stimulus was present, but he was not particularly sure so he would only bet a small amount of money on it. Usually the bet options consist of a given set, for example 0.20, 0.40 and 0.60 cents. In this way, a post-decision wagering scale can be made. The participant’s choice on the scale resembles their subjective awareness. For example, if a participant bets 0.60 cents on the target being present this would indicate that the participant was completely aware of the target being there. On the other hand, in case a participant bets only 0.20 on the target being present this would mean that the participant was not fully aware of the target but probably did take in some of its information, meaning the participant was in an “in-between” state of awareness. In this way, it is expected that if participants only use the extreme ends of the scale visual awareness would be all-or-none, and if participants also regularly use the intermediate points of the scale visual awareness would be gradual. In a study of Nieuwenhuis and de Kleijn (2011) using post-decision wagering, they found a rather uniform rating pattern, thus concluding consciousness to be gradual.

However, a critique on this measure is that wagering is susceptible to certain psychological factors irrelevant to the emergence of consciousness, such as the

participant’s gambling strategy (Dienes & Seth, 2010; Nieuwenhuis & de Kleijn, 2011; Clifford, Arabzadeh & Harris, 2006). Besides, the wagering offer could come about by implicitly learned gambling behavior (Nieuwenhuis & de Kleijn, 2011). When taking this into account there is a possibility that using post-decision wagering participant’s responses cannot reliably lead to conclusions about the emergence of consciousness.

To summarize, the above-mentioned measures can be used to give an

indication of the nature of the emergence of consciousness. When participants tend to only choose the extreme ends of a given scale, independent of the measure used, this would suggest the emergence of consciousness to be all-or-none. Participants would state that their subjective awareness consists only of two discrete options: seeing the stimulus or not seeing the stimulus. On the other hand, when participants likewise choose the intermediate points of a given scale, for example choosing the “brief

5 glimpse” option on the Perceptual Awareness Scale or the 0.40 cents bet option on the post-decision wagering scale, this would suggest the emergence of consciousness to be gradual. In that case participants would state that their subjective awareness consists of multiple options that develop in a gradual manner: from not seeing the stimulus at all to getting a small notion of the stimulus to having a better notion of the stimulus to eventually gaining full conscious access to the stimulus.

Hence, by looking at previous studies, when confidence ratings, Perceptual Awareness Scale and post-decision wagering were used, the emergence of

consciousness seemed to be gradual instead of all-or-none (Overgaard, Rote, Mouridsen & Ramsoy, 2006; Nieuwenhuis & de Kleijn, 2011). However, when a continuous scale was used, the emergence of consciousness seemed to be all-or-none (Sergent & Dehaene, 2004b; Del Cul, Baillet & Dehaene, 2007). These findings show that considering the emergence of consciousness as an all-or-none or gradual

phenomenon depends on the measure that is used to establish subjective awareness. Moreover, studies that compared these measures have found contrasting results when looking at the differences in subjective awareness ratings (Sandberg,

Timmermans, Overgaard & Cleeremans, 2010; Szczepanowski, Traczyk, Wierzchón & Cleeremans, 2013).

Furthermore, Windey, Gevers and Cleeremans (2013) proposed that the emergence of consciousness differs across tasks and that this difference can be attributed to the level of processing that a task requires. They claim that low-level processing tasks cause awareness to be more graded and that high-level processing tasks cause awareness to be more dichotomous. Anzulewicz et al. (2015) and

Windey, Vermeiren, Atas and Cleeremans (2014) carried out experiments that provided support for these claims. These findings suggest that there could be more factors involved that may have some effect on the outcome of subjective awareness ratings, in the sense of them being gradual or all-or-none.

On top of this, an important critique regarding the studies that used subjective awareness measures is that in all these different studies often different

psychophysical techniques have been used to measure a stimulus gaining access to awareness. For example, Sergent and Dehaene (2004b) only found a bimodal rating pattern on their subjective awareness measure when using the attentional blink, and not when using backward masking. This poses the question whether these

differences in subjective awareness ratings could, apart from measure or task, also be caused by the psychophysical technique that is being used.

A frequently used technique in consciousness research is the attentional blink (AB) (Broadbent & Broadbent, 1987; Shapiro, Arnell & Raymond, 1997). In this

6 technique two masked targets (T1 and T2) are presented with changeable time gaps between the two targets. The attentional blink effect is a phenomenon that occurs when T2 is presented approximately between 200 and 500 ms after T1, which causes the participant to be unable to detect T2, despite of detecting T1. When the

participant is instructed to ignore T1, or when the time gap between T1 and T2 is larger than 500 ms, the attentional blink effect does not arise. By using this technique, a stimulus gaining access to consciousness can be measured. It is expected that with an interval between 200 and 500 ms T2 will not reach full awareness and with a smaller or larger interval T2 will. In this way, by comparing different intervals and combining this technique with a subjective awareness measure the emergence of consciousness can be examined.

Another often-used technique regarding consciousness research is backward masking (BM) (Breitmeyer & Ogmen, 2000). In this technique, a target is presented and immediately followed by a series of masking stimuli. The time that the target is shown before the masking stimuli are presented (presentation duration) can differ, which will cause the target to reach full awareness at longer presentation durations. On the other hand, at shorter presentation durations the target will not reach full awareness. In this way, by comparing different presentation durations and by combining this technique with a subjective awareness measure the emergence of consciousness can be investigated.

Lastly, a regularly used technique in consciousness research is continuous flash suppression (CFS) (Tsuchiya & Koch, 2005). In this technique, a stationary target is presented by slowly fading in to one eye, whereas random flashed patterns are presented to the other eye. Normally the flashed patterns cause the target to be suppressed and inaccessible to awareness for up to several seconds. By using different presentation durations, the target will sometimes fail to reach full awareness. In this way, by comparing different presentation durations and

combining this with a subjective awareness measure, the emergence of consciousness can be examined.

In the breaking-continuous flash suppression paradigm (b-CFS) different stimuli are compared based on the time it takes them to gain access to awareness (Stein & Sterzer, 2014). Usually participants perform some sort of detection or localization task during continuous flash suppression so that there is an objective measure to compare the stimuli on. In this way differences in for example the detection time of stimuli can infer to differences in unconscious processing across stimuli. Gayet, Van der Stigchel and Paffen (2014) provided a review on this technique and showed that multiple studies using b-CFS found that differences in

7 unconscious processing of specific types of stimuli can be detected. An example of a reliable effect that has been found using b-CFS is the inversion effect, in which upright faces gain access to awareness faster than inverted faces (Jiang, Costello & He, 2007; Zhou, Zhang, Liu, Yang & Qu, 2010; Stein, Sterzer & Peelen, 2012).

In the present study, the influence of different psychophysical techniques on the emergence of visual awareness will be examined. To test whether psychophysical technique has an effect on the emergence of visual awareness, three techniques will be used: AB, BM and b-CFS. When taking the findings of Sergent and Dehaene (2004b) into account, it is expected that when using AB the subjective awareness ratings will turn out to be more bimodal and thus suggesting all-or-none awareness, than when using BM. Furthermore, it is expected that when b-CFS is used the

subjective awareness ratings will also turn out to be bimodal and therefore indicating none awareness. This expectation originates from the idea that for b-CFS all-or-none awareness seems to be implicitly assumed. Many studies that use b-CFS use response time measures to track access to awareness, which is therefore assumed to occur at one particular point in time (Gayet, Van der Stigchel & Paffen, 2014). These three techniques will be combined with the Perceptual Awareness Scale as a

subjective awareness measure and a simple localization task as an objective performance measure. Consequently, the resulting distribution of subjective awareness ratings and localization accuracy scores will be compared across the different techniques.

8

Methods

The current study consisted of two experiments, one in which participants

performed a task using both AB and BM, and one in which participants performed a task using only b-CFS.

Experiment 1: Attentional Blink and Backward Masking

Participants

Participants took part in this experiment by registering themselves on

www.lab.uva.nl and did so in return for research credits (1.5 credits). The experiment consisted of 10 participants in total (age ranging from 19 to 27; 4 female; 8 right

handed). All participants said to have normal or corrected-to-normal vision.

Design and Stimuli

All participants took part in two sessions, one for each technique (AB and BM). For all participants objective performance (localization task) and subjective performance (awareness ratings) were measured.

For both AB and BM the task and stimuli were largely the same. The stimuli were presented on a screen with a resolution of 100 Hz. In all cases the critical target stimuli consisted of 16 upright (Figure 1) and 16 inverted (Figure 2) faces and the task was to localize this target. The outer facial features were cut and the faces were

adjusted to match the grey background. The size of the face targets was 100 x 100 pixels. Upright and inverted faces were used as a target the same number of trials and the specific faces (out of the 16 options) were also used an equally number of trials. This was done in a randomized order. The experiment consisted of 384 trials in total, per technique used. After every 128 trials, a break was included, with a

duration of minimally 10 seconds. Thus, each session consisted of two breaks.

9 Attentional Blink

For AB (Figure 6), three streams of rapid serial visual presentations of masking stimuli were presented. The masking stimuli were a set of scrambled faces (16 upright and 16 inverted) (Figure 3). At a given moment T1 was presented in the middle stream (green circles (Figure 4) or diamonds (Figure 5) in the shape of a face) and sometime later in the two streams surrounding the middle stream T2

(upright/inverted face) was presented at one of the two locations (left/right). We used three time intervals between T1 and T2, lag 2, lag 3 and lag 8. The AB effect would be expected to be present at the shorter lags. All three lags occurred equally often and in a randomized order.

Figure 3. Example of a scrambled face mask. Figure 4. Example of a circles T1. Figure 5. Example of a diamonds T1.

Figure 6. Attentional Blink paradigm.

Backward Masking

For BW (Figure 7), a squared background was used to present the target and masking stimuli on. In this paradigm the masking stimuli were Mondrian-like

masks. The target was presented at either the left or right side of the background and followed by the masking stimuli. We used three different presentation durations

10 before the onset of the masking stimuli: 10 ms, 20 ms and 20 ms with blank. All three presentation durations occurred equally often and in a randomized order.

Figure 7. Backward Masking paradigm. Perceptual Awareness Scale

The Perceptual Awareness Scale (PAS) was used to obtain subjective awareness ratings. The numbers 1 through 4 were displayed on the screen, each containing a label concerning the visibility of the shown target. In this experiment the traditional PAS was used, with labeling as follows: 1 = “no experience”, 2 = “brief glimpse”, 3 = “almost clear experience”, 4 = “clear experience”.

11

Procedure

The experiment took place at the lab of the UvA in Amsterdam. Participation took 1.5 hours. At the beginning of the experiment, participants read and signed an informed consent, agreeing to participate in the experiment. Participants started off with one of the two techniques with the use of counterbalancing. At first participants read an instruction form of the experiment according to the corresponding technique. Before the real testing phase began participants first did some practice trials to see if they could perform the task correctly and during this they were free to ask questions. Right before the start of the experiment participants were asked to provide some general information about themselves: age, gender and handedness.

In the testing phase participants were instructed to look at the screen and search for a face (with the note that the faces could also be inverted). The instructions were slightly different for the two techniques (e.g. for AB the instructions asked to pay attention to T1). At the end of each trial, participants were asked to localize the target (by pressing the left or right arrow on the keyboard) and to rate their

subjective awareness of the target (by pressing the keys 1, 2, 3 or 4, according to the corresponding label). In the AB session participants were also asked to first identify T1 and feedback was provided on this by presenting a green (correct) or red

(incorrect) fixation cross, to illustrate the importance of detecting T1.

After the first session was completed, participants immediately took part in a second session with the other technique.

Experiment 2: breaking-Continuous Flash Suppression

Participants

Participants took part in this experiment by registering themselves on

www.lab.uva.nl and did so in return for research credits (1.5 credits). The experiment consisted of 7 participants in total (age ranging from 21 to 68; 4 female; 6 right

handed). All participants said to have normal or corrected-to-normal vision.

Design and Stimuli

Experiment 2 was almost identical to experiment 1, except here b-CFS was the only technique used. For all participants objective performance (localization task) and subjective performance (awareness ratings) were measured. The subjective

awareness measure was implemented in the same way as in experiment 1. The task and stimuli were exactly the same as in experiment 1. Only a few small changes were made to make the task fit with b-CFS. The experiment consisted of 640 trials in total.

12 After every 128 trials, a break was included, with a duration of minimally 10 seconds. Thus, there were four breaks.

For b-CFS, by the use of a mirror stereoscope, a target was presented to one eye, slowly gaining in contrast over 200 ms, while masking stimuli were presented to the other eye. The target was always presented to the dominant eye. The masking stimuli consisted of Mondrian-like masks with a refresh rate of 10 Hz plus three backward masks at the end of each trial. The target was presented at either the left or right side of the background. We used five different presentation durations: 500 ms, 750 ms, 1250 ms, 2000 ms and 3000 ms. All five presentation durations occurred equally often and in a randomized order.

Figure 9. Breaking-Continuous Flash Suppression paradigm.

Procedure

The experiment took place at the lab of the UvA in Amsterdam. Participation took 1.5 hours. At the beginning of the experiment, participants read and signed an informed consent, agreeing to participate in the experiment. Then the dominant eye of the participant was determined with a simple visual test. After that participants read an instruction form of the experiment. Before the real testing phase began participants first did some practice trials to see if they could perform the task correctly and during this they were free to ask questions. Right before the start of the experiment participants were asked to provide some general information about themselves: age, gender and handedness.

In the testing phase participants were instructed to look at the screen and search for a face (with the note that the faces could also be inverted). At the end of each trial, participants were asked to localize the target (by pressing the left or right arrow on the keyboard) and to rate their subjective awareness of the target (by pressing the keys 1, 2, 3 or 4, according to the corresponding label).

13

Data Analysis Plan

First, for each individual technique the mean PAS rating, mean target localization accuracy and the distribution of PAS ratings will be compared across the different presentation durations. This will be done by the use of One-Way Repeated Measures ANOVAs and Factorial Mixed ANOVAs. Then at last a combined analysis will be performed in which the distribution of PAS ratings will be compared across presentation durations and across the different techniques all in once. This will be done by the use of a Factorial Mixed ANOVA. For all within-subject variables the assumption of sphericity will be checked with the use of Mauchly’s Test. If the assumption of sphericity is violated, a Greenhouse-Geisser correction will be used and the adjusted degrees of freedom will be reported.

14

Results

Attentional Blink

At first the few trials in which T1 was not accurately identified were discarded from the AB dataset. Only trials in which T1 was accurately identified were included in the analysis. T1 was correctly identified with a mean of 0.9729 and a standard deviation of 0.0131.

Subjective Awareness Ratings

A One-Way Repeated Measures ANOVA was performed to test whether mean PAS ratings differed across different time intervals between T1 and T2. Therefore mean PAS rating was used as a dependent variable and time interval between T1 and T2 (three levels: lag 2, lag 3 and lag 8) as an independent variable. There was a main effect of time interval between T1 and T2 on mean PAS rating (F(1.263, 11.369) = 10.451, p = 0.005). Figure 10 shows that when lag increases, mean PAS rating also increases.

15

Objective Performance

Another One-Way Repeated Measures ANOVA was performed to test whether mean target localization accuracy differed across different time intervals between T1 and T2. Here mean target localization accuracy was used as a dependent variable and time interval between T1 and T2 (three levels: lag 2, lag 3 and lag 8) as an

independent variable. There was a main effect of time interval between T1 and T2 on mean target localization accuracy (F(1.170, 10.528) = 6.505, p = .024). Figure 11 shows that as lag increases, mean target localization accuracy also increases.

Figure 11. Mean target localization accuracy across different time intervals between T1 and T2.

Distribution of Subjective Awareness Ratings

At last another One-Way Repeated Measures ANOVA was performed to test whether the distribution of PAS ratings differed across different time intervals between T1 and T2. For this the proportion of PAS ratings was used as a dependent variable and time interval between T1 and T2 (three levels: lag 2, lag 3 and lag 8) and PAS rating (four levels: 1, 2, 3 and 4) as independent variables. A main effect of PAS rating on the proportion of PAS ratings (F(1.490, 13.409) = 18.001, p < .001) was

present. There also was an interaction effect between time interval and PAS rating on the proportion of PAS ratings (F(1.500, 13.498) = 8.634, p = .006).

Figure 12 shows the course of this interaction effect. As the lag increases, the proportion of PAS rating 1 decreases, while the proportion of PAS rating 2 increases.

16 Moreover, the proportion of PAS rating 3 and 4 stay roughly the same across all three lags. By looking at Figure 12 the distribution of PAS ratings can be best described as skewed to the right, with peaks at PAS rating 1 and 2. PAS rating 3 and 4 were also selected, but never with a proportion higher than 0.15. This trend applies to all three lags.

Figure 12. Distribution of PAS ratings across different time intervals between T1 and T2.

Backward Masking

Subjective Awareness Ratings

Firstly, a One-Way Repeated Measures ANOVA was performed to test whether mean PAS ratings differed across different presentation durations. To accomplish this mean PAS rating was used as a dependent variable and presentation duration (three levels: 10 ms, 20 ms and 20 ms with blank) as an independent variable. There was a main effect of presentation duration on mean PAS rating (F(1.144, 10.299) = 38.057, p < .001). Figure 13 shows that when lag increases, mean PAS rating also increases.

17 Figure 13. Mean PAS ratings across different presentation durations.

Objective Performance

Another One-Way Repeated Measures ANOVA was performed to test whether mean target localization accuracy differed across different presentation durations. In this analysis mean target localization accuracy was used as a dependent variable and presentation duration (three levels: 10 ms, 20 ms and 20 ms with blank) as an independent variable. A main effect of presentation duration on mean target

localization accuracy was present (F(2, 18) = 43.713, p < .001). Figure 14 displays that as lag increases, mean target localization accuracy also increases.

18 Figure 14. Mean target localization accuracy across different presentation durations.

Distribution of Subjective Awareness Ratings

Finally, a One-Way Repeated Measures ANOVA was performed to test whether the distribution of PAS ratings differed across different presentation durations. For this the proportion of PAS ratings was used as a dependent variable and presentation duration (three levels: 10 ms, 20 ms and 20 ms with blank) and PAS rating (four levels: 1, 2, 3 and 4) as independent variables. There was a main effect of PAS rating on the proportion of PAS ratings (F(1.657, 14.915) = 12.451, p = .001). There also was an interaction effect between presentation duration and PAS rating on the proportion of PAS ratings (F(2.284, 20.560 = 17.386, p < .001).

Figure 15 displays the course of this interaction effect. As presentation

duration increases, the proportion of PAS rating 1 decreases, while the proportion of PAS rating 2 increases. The proportion of PAS rating 3 and 4 also increase when presentation duration increases, but more slightly than as with PAS rating 2. When looking at Figure 15 the distribution of PAS ratings can be best described as skewed to the right, with peaks at PAS rating 1 and 2. PAS rating 3 and 4 were also selected, but noticeably less than PAS rating 1 and 2. This applies to all three presentation durations.

19 Figure 15. Distribution of PAS ratings across different presentation durations.

Breaking-Continuous Flash Suppression

At first two participants were excluded from the b-CFS dataset due to ceiling effects. These two participants correctly localized the target in respectively 99.6% and 99.0% of all trials.

Subjective Awareness Ratings

A One-Way Repeated Measures ANOVA was performed to test whether mean PAS ratings differed across different presentation durations. For this mean PAS rating was used as a dependent variable and presentation duration (five levels: 500 ms, 750 ms, 1250 ms, 2000 ms and 3000 ms) as an independent variable. A main effect of presentation duration on mean PAS rating was present (F(1.160, 4.639) = 13.962, p = .014). Figure 16 shows that as lag increases, mean PAS rating also increases.

20 Figure 16. Mean PAS ratings across different presentation durations.

Objective Performance

Another One-Way Repeated Measures ANOVA was performed to test whether mean target localization accuracy differed across different presentation durations. For this mean target localization accuracy was used as a dependent variable and presentation duration (five levels: 500 ms, 750 ms, 1250 ms, 2000 ms and 3000 ms) as an

independent variable. There was a main effect of presentation duration on mean target localization accuracy was present (F(4, 16) = 20.117, p = .045). Figure 17 displays that when lag increases, mean target localization accuracy also increases.

21 Figure 17. Mean target localization accuracy across different presentation durations.

Distribution of Subjective Awareness Ratings

Lastly, a One-Way Repeated Measures ANOVA was performed to test whether the distribution of PAS ratings differed across different presentation durations. For this analysis the proportion of PAS ratings was used as a dependent variable and

presentation duration (five levels: 500 ms, 750 ms, 1250 ms, 2000 ms and 3000 ms) and PAS rating (four levels: 1, 2, 3 and 4) as independent variables. There was a main effect of PAS rating on the proportion of PAS ratings (F(1.143, 5.723) = 11.428, p = .013). There also was an interaction effect between presentation duration and PAS rating on the proportion of PAS ratings (F(1.926, 7.703) = 9.518, p = .009).

Figure 18 shows the course of this interaction effect. As presentation duration increases, the proportion of PAS rating 1 decreases, while the proportion of PAS rating 4 increases. The presentation duration 2000 ms seems to be an important turning point in the distribution of PAS rating 1 and 4. The proportion of PAS rating 2 and 3 stay roughly the same across all five presentation durations, except for a small increase at a 1250 ms presentation duration for PAS rating 2. When looking at Figure 18 the distribution of PAS ratings can be best described as skewed to the right, with peaks at PAS rating 1 and 2. However, for the longest presentation durations (2000 ms and 3000 ms) all PAS ratings except for PAS rating 3 were almost equally often selected.

22 Figure 18. Distribution of PAS ratings across different presentation durations.

Across Techniques

To be able to compare the distribution of PAS ratings across the different techniques by also looking at the different presentation durations, the levels of presentation duration of b-CFS had to be narrowed down to three. To do this, the mean PAS ratings and mean target localization accuracy across presentation durations for AB and BM were compared to those for b-CFS (Figure 19). The three b-CFS presentation durations that best matched the means of PAS ratings and target localization

accuracy of the other techniques were selected. Therefore, the presentation durations 500 ms, 750 ms and 1250 ms for b-CFS were included in the next analysis.

Presentation duration

AB BM b-CFS

Mean PAS rating 1 1.504 1.356 1.284

2 1.577 1.595 1.490

3 1.733 2.066 1.790

4 - - 2.004

5 - - 2.160

Mean target localization accuracy 1 0.679 0.674 0.622

2 0.751 0.794 0.694

3 0.799 0.900 0.758

4 - - 0.816

5 - - 0.862

Figure 19. Mean PAS rating and mean target localization accuracy across presentation durations for AB, BM and b-CFS. To test whether the distribution of PAS ratings across different presentation durations differed across techniques a Factorial Mixed ANOVA was performed. For

23 this analysis technique (three levels: AB, BM and b-CFS) was used as a between-subjects independent variable and presentation duration (three levels: short, middle, long) and PAS ratings (four levels: 1, 2, 3 and 4) were used as within-subjects

independent variables. The proportion of PAS ratings was used as a dependent variable.

There was a main effect of PAS rating (F(1.605, 35.305) = 41.098, p < .001) but no interaction effect between PAS rating and technique (F(6, 66) = .201, p = .975) on the proportion of PAS ratings. Furthermore, an interaction effect between

presentation duration and PAS rating (F(2.482, 54.605) = 26.811, p < .001) and an interaction effect between presentation duration, PAS rating and technique (F(12, 132) = 3.192, p < .001) on the proportion of PAS ratings were present.

Due to the lack of statistical power, since the b-CFS data only consisted of 5 participants, no reliable further analysis could be performed on the interaction effect between presentation duration, PAS rating and technique. When looking at this interaction effect with the use of a Factorial Mixed ANOVA there is an interaction effect between presentation duration and technique for PAS rating 1 (F(4, 44) = 5.977,

p = .001), and not for PAS rating 2 (F(4, 44) = 1.807, p = .145), 3 (F(4, 44) = 2.235, p =

.081) or 4 (F(4, 44) = .747, p = .565). Figure 20 shows this interaction effect between presentation duration and technique for PAS rating 1. A possible explanation for this effect is that the decrease in PAS rating 1 across presentation durations (especially from middle to long presentation duration) seems to be much greater for BM than for AB and b-CFS.

24 Figure 20. Proportion of PAS rating 1 across presentation durations and techniques.

25

Discussion

In the current study, the influence of psychophysical technique on the emergence of visual awareness was examined. When looking at the results it seems that for all techniques used the emergence of visual awareness was more gradual than all-or-none. For all techniques used the distribution of subjective awareness ratings could be best described as skewed to the lower visibility ratings. This means that

participants overall had trouble with consciously perceiving the target. However, participants frequently selected all the different scale points of the PAS, especially for the longer presentation durations, thus indicating that participants were both

partially and fully aware of the face target for some trials. This finding suggests that the emergence of visual awareness is gradual and not all-or-none, regardless of the technique used.

Evidently, the results of this study do not support the global neuronal

workspace theory (Sergent & Dehaene, 2004b), as participants commonly reported to be partially aware of the target. However, one could argue that when the “brief glimpse” option on the PAS was chosen, this would have meant that the target had already reached full awareness. When looking at it this way, PAS rating 1 indicates unawareness and PAS rating 2 (together with 3 and 4) indicates awareness. If this would be the case, the results of this study would not necessarily rule out all-or-none awareness. This critique shows how complex the concepts regarding consciousness research really are, and how open they are to interpretation.

When comparing the present findings to those of previous studies, only the findings regarding BM are in line with the previous literature, since Sergent and Dehaene (2004b) did not find a bimodal rating pattern when using BM. For AB, the distribution of subjective awareness ratings was clearly not bimodal, opposed to the study of Sergent and Dehaene (2004b). For b-CFS, the distribution of subjective awareness ratings also did not correspond with the implicitly assumed notion of all-or-none awareness when using this paradigm (Gayet, Van der Stigchel & Paffen, 2014).

Nevertheless, there are some shortcomings to this study which will be

discussed next. Firstly, the use of the PAS as the only subjective awareness measure. As stated in the introduction, the outcome of studies concerning the emergence of consciousness heavily depends on the subjective awareness measure that is being used. The use of the PAS as a subjective awareness measure already had an influence on the findings of this study. For example, a before mentioned critique on the PAS is

26 that the use of this scale can lead to scale bias, in which participants feel obligated to make use of all the given scale-points, regardless of their true subjective awareness (Windey, Vermeiren, Atas & Cleeremans, 2014). To overcome this complication, it would be really useful to include at least one other subjective awareness measure (such as a continuous scale) so that possible differences in the emergence of visual awareness could also be related to the measure that was being used.

Secondly, the differences between the presentation durations across

techniques. As a result of each technique having different demands when it comes to ascribing appropriate presentation durations, the used presentation durations are rather different for each technique. In this study presentation duration was an important factor, showing how the distribution of subjective awareness ratings can differ according to the level of difficulty of a task. In the combined techniques analysis, the distribution of subjective awareness ratings was compared across presentation durations. When comparing this across techniques, it is essential that the presentation durations for each individual technique are also strictly comparable with the presentation durations of the other techniques. This was not exactly the case. For starters b-CFS consisted of two more presentation durations than the other techniques. Apart from that, it is not certain that for example the lag 2 for AB, 10 ms for BM and 500 ms for b-CFS all resemble the same level of task difficulty. This problem is hard to solve, but one could for example do a pilot study in which

different presentation durations are compared on mean subjective awareness ratings and mean accuracy scores, after which the best matching presentation durations will be used for the main study. In the present study, this comparison was also made, but in a less thorough manner.

Lastly, the combination of a within- and between-subjects design. Due to both the low sample size for the analysis (N = 10 and N = 5) and the fact that the

techniques AB and BM were used on the same participants, while b-CFS was used on different participants, the results of AB and BM are more difficult to relate to b-CFS than to each other. It is possible that this combination of low sample size and a different participant group for b-CFS has made the results for b-CFS look differently than they would when b-CFS was used on the same group of participants. A way to solve this limitation is to either use all techniques on the same group of participants, or to use all techniques on a new different group of participants, thus having an experimental design that is either fully within-subjects or fully between-subjects.

Apart from these shortcomings, the current study does provide some

interesting results regarding the emergence of visual awareness. No real conclusions can be made because of the explorative nature of this study. Nonetheless, the present

27 findings seem to agree with other studies such as those of Overgaard, Rote,

Mouridsen and Ramsoy (2006) and Nieuwenhuis and de Kleijn (2011), in that the emergence of visual awareness is more gradual than all-or-none. A suggestion for follow-up research is to include multiple subjective awareness measures in the design to see whether this causes differences in the distribution of subjective

awareness ratings across different techniques. Altogether, this study points out that the emergence of visual awareness is most likely to be more gradual than all-or-none, and that psychophysical technique does not seem to have an influence on this

28

References

Anzulewicz, A., Asanowicz, D., Windey, B., Paulewicz, B., Wierzchoń, M., & Cleeremans, A. (2015). Does level of processing affect the transition from unconscious to conscious perception? Consciousness and Cognition, 36, 1–11. Asplund, C. L., Fougnie, D., Zughni, S., Martin, J. W., & Marois, R. (2014). The

attentional blink reveals the probabilistic nature of discrete conscious perception. Psychological Science, 25(3), 824–831.

Bar, M., Tootell, R. B. H., Schacter, D. L., Greve, D. N., Fischl, B., Mendola, J. D., Rosen, B. R., & Dale, A. M. (2001). Cortical mechanisms specific to explicit visual object recognition. Neuron, 29(2), 529–535.

Breitmeyer, B. G., & Ogmen, H. (2000). Recent models and findings in visual backward masking: A comparison, review, and update. Perception &

Psychophysics, 62(8), 1572–1595.

Broadbent, D. E., & Broadbent, M. H. P. (1987). From detection to identification: Response to multiple targets in rapid serial visual presentation. Perception &

Psychophysics, 42(2), 105–113.

Clifford, C. W. G., Arabzadeh, E., & Harris, J. A. (2008). Getting technical about awareness. Trends in Cognitive Sciences, 12(2), 54–58.

Del Cul, A., Baillet, S., & Dehaene, S. (2007). Brain dynamics underlying the

nonlinear threshold for access to consciousness. PLoS Biology, 5(10), 2408–2423. Dienes, Z., & Seth, A. (2010). Gambling on the unconscious: A comparison of

wagering and confidence ratings as measures of awareness in an artificial grammar task. Consciousness and Cognition, 19(2), 674–681.

Gayet, S., Van Der Stigchel, S., & Paffen, C. L. E. (2014). Breaking continuous flash suppression: Competing for consciousness on the pre-semantic battlefield.

Frontiers in Psychology, 5, 1–10.

Nieuwenhuis, S., & de Kleijn, R. (2011). Consciousness of targets during the attentional blink: A gradual or all-or-none dimension? Attention, Perception &

Psychophysics, 73, 364–373.

Overgaard, M., Rote, J., Mouridsen, K., & Ramsøy, T. Z. (2006). Is conscious perception gradual or dichotomous? A comparison of report methodologies during a visual task. Consciousness and Cognition, 15(4), 700–708.

Persaud, N., McLeod, P., & Cowey, A. (2007). Post-decision wagering objectively measures awareness. Nature Neuroscience, 10(2), 257–261.

29

Phenomenology and the Cognitive Sciences, 3(1), 1–23.

Sandberg, K., Timmermans, B., Overgaard, M., & Cleeremans, A. (2010). Measuring consciousness: Is one measure better than the other? Consciousness and Cognition,

19(4), 1069–1078.

Sergent, C., & Dehaene, S. (2004a). Neural processes underlying consious perception: Experimental findings and a global neural workspace framework. Journal of

Physiology - Paris, 98, 374-384.

Sergent, C., & Dehaene, S. (2004b). Is consciousness a gradual phenomenon?

Psychological Science, 15(11), 720–728.

Shapiro, K. L., Arnell, K. M., & Raymond, J. E. (1997). The attentional blink. Trends in

Cognitive Sciences, 1(8), 291-296.

Stein, T., & Sterzer, P. (2014). Unconscious processing under interocular suppression: Getting the right measure. Frontiers in Psychology, 5, 1–5.

Stein, T., Sterzer, P., & Peelen, M. V. (2012). Privileged detection of conspecifics: Evidence from inversion effects during continuous flash suppression. Cognition,

125(1), 64–79.

Jiang, Y., Costello, P., & He, S. (2007). Processing of invisible stimuli: Advantage of upright faces and recognizable words in overcoming interocular suppression.

Psychological Science, 18(4), 349–355.

Szczepanowski, R., Traczyk, J., Wierzchón, M., & Cleeremans, A. (2013). The perception of visual emotion: Comparing different measures of awareness.

Consciousness and Cognition, 22(1), 212–220.

Tsuchiya, N., & Koch, C. (2005). Continuous flash suppression reduces negative afterimages. Nature Neuroscience, 8(8), 1096–1101.

Vul, E., Hanus, D., & Kanwisher, N. (2009). Attention as inference: Selection is probabilistic; responses are all-or-none samples. Journal of Experimental

Psychology: General, 138(4), 546–560.

Windey, B., Gevers, W., & Cleeremans, A. (2013). Subjective visibility depends on level of processing. Cognition, 129(2), 404–409.

Windey, B., Vermeiren, A., Atas, A., & Cleeremans, A. (2014). The graded and dichotomous nature of visual awareness. Philosophical Transactions of the Royal

Society of London. Series B, Biological Sciences, 369(1641), 1–11.

Zhou, G., Zhang, L., Liu, J., Yang, J., & Qu, Z. (2010). Specificity of face processing without awareness. Consciousness and Cognition, 19(1), 408–412.