The Role of Working Memory in the Attentional Blink
Mels Boerkamp University of Amsterdam
Begeleidster: Daniel Lindh
Studentnummer: 10023151
Aantal woorden abstract: 138
Abstract
Attentional blink is the brief period of inattention that occurs shortly after a Rapid Serial Visual Presentation (RSVP) of images. This study examined the role of working memory when an attentional blink (AB) was present. Previous research suggest that working memory is related to the size of the attentional blink because of a better efficiency of encoding information and to lag-1 sparing, the combined processing of multiple stimuli within a single attentional period. In the study 22 participants did an attentional blink task and an n-back task. In all participants an attentional blink was present. Working memory had a positive effect on the accuracy on the first lag, possibly due to lag-1 sparing. Second, no link was found between the magnitude of the AB and working memory. Limitations of the study and suggestions for future research are discussed.
Science has come a long way in the pursuit of analysing the path of visual processing in the human brain. From the retina visual stimuli is transformed into electrical signals and send to the visual cortex. Low-level features such as edges and corners are primarily analysed in the early visual cortex. Further down the ventral stream progressively complex high-level features such as objects, scenes and integration of multiple aspects of an image are analysed. Research on primates suggest that the tangling of multiple aspects of an object is the mechanism behind recognition and is stable with changes for example with orientation of an image (Dicarlo & Cox, 2007). But when images of objects were transformed to preserve their mid-level features but were made unrecognisable, it still elicited a broad activation in object-selective brain regions along the ventral pathway (Long, Yu & Konkle, 2017). Although it’s known which areas relate to certain visual stimuli there is still no consensus on how this input is processed precisely and eventually comes into our awareness. Broadbent and Broadbent (1987) discovered that when subjects are presented with a Rapid Serial Visual Presentation (RSVP) of images and are instructed to identify two targets (T1 and T2, respectively) embedded in a series of distractors, participants showed a decreased ability to report the T2 when presented 200-500ms post T1. Raymond, Shapiro and Arnell (1992) showed that this effect is only present when subjects are instructed to attend the first target, linking the phenomena to a depletion of attentional resources and thus named it the Attentional Blink (AB). The closing of an attentional gate increases the chance that new stimuli will be missed, therefore the accuracy during and after this period are often compared. The difference in attentional resources during lag 2 and lag 8 (200ms and 800ms post T1 respectively) has also been investigated by looking at event-related potentials (ERP’s) of EEG recordings. Here, the P300-wave, event-related to attentional processes (Gray, Ambady, Lowenthal & Deldin, 2004), is reduced during lag 2 compared to lag 8 (Vogel & Luck, 2002). There have been countless of models describing the AB (Dux & Marois, 2009), where most theories rely on a division between pre and post-perceptual processes. The
earliest account of this sort is the two-stage model (Chun & Potter, 1995). Here, both targets are processed up to a high level of semantic representation in the first stage, where representations are fragile and subject to decay and overwriting of distracting stimuli. This is followed by the second stage, consisting of a capacity-limited attention processing stage, where visual representations are brought up to a more durable working memory state. This first stage was corroborated by electrophysiological evidence (Luck, Vogel & Shapiro, 1996) that meanings of words can be accessed even during an attentional blink, when participants weren’t able to report this 2 seconds later. Second, there was evidence for a rapid feature analysis as proposed in the second stage (Evans and Treisman, 2005), but it was found that for identification and localization of objects attention is required. Furthermore an fMRI-study on the attentional blink used images of scenery that activated brain regions involved in scene representations even when these images were not consciously perceived (Marois & Chun, 2004). More importantly, there was evidence found for an attentional bottleneck, the frontal cortex was only activated when scenes were correctly reported.
However, an attentional blink doesn’t seem to be universal. Martens, Munneke, Smid and Johnson (2006) has shown that not everyone has an attentional blink. These nonblinkers showed no or a very small attentional blink and seem to have less difficulty rejecting distractors than blinkers and therefore report targets more accurately (Martens and Valchev, 2009). These results would imply that the size of the attentional blink would be correlated to the inhibition of distractors or the efficiency of encoding information. This efficiency of encoding information was found to be related to the visual short term memory by Vogel, mcCollough and Machizawa (2005). Individuals with a high capacity visual short term memory differed only in the efficiency of encoding relevant items compared to low capacity individuals, who also encode and maintain information on irrelevant items. If the encoding of information is linked to the attentional blink, then similarly, the size of the attentional blink could be predicted by the
number of items one could hold into working memory. Research on the operation span of working memory found exactly this (Colzato, Spapé, Pannebakker and Hommel, 2007). It is therefore hypothesized that the operation span of working memory was negatively correlated to the size of the attentional blink. The closing of the attentional gate is supported by the existence of lag-1 sparing. Lag-1 sparing occurs when items come 100 ms after each other. Due to this close temporal proximity both items are transferred from short-term memory to working memory where it is possible to become aware of it. Results from research of Hommel and Akyürek (2005) on lag-1 sparing suggest that items compete for attention. If a capacity-limited bottleneck is indeed the mechanism behind the attentional blink, than this could be related to the operation span of working memory. Better and more efficient encoding could lead to more lag-1 sparing. It could be suggested from these results that the attentional blink is modulated by suppressing distractors due to a more efficient working memory. The link between suppressing irrelevant information and the size of the attentional blink seems plausible, although not all research into these areas could find a relation between one another (Martens & Johnson, 2009).
Based on earlier research (Vogel, mcCollough and Machizawa, 2005; Colzato, Spapé, Pannebakker and Hommel, 2007) we will be conducting an experiment with an attentional blink task and a working memory task, to investigate the possible relation between the attentional blink and working memory. Second, we will examine if working memory is associated with the lag-1 sparing. It is expected that working memory will be negatively related to the size of the attentional blink as was found by Colzato and colleagues (2007).
Method
Participants
Twenty-two participants (18-26 years old, 17 females) were recruited via online media through University of Amsterdam Psychology-groups on Facebook and on the website of the university’s experiments-section where students could apply in exchange for money (€20 per session) or student credits. All participants had normal, or corrected to normal, vision. All participants signed an informed consent form overseen by the university’s ethics board.
Material
40 black and white images, classified in 8 different categories were used. All images were used in two different tasks: an attentional blink task (as used in Evans & Treisman, 2005) and an n-back task (Carlson et al., 1998). Categories ranged from animate (bears, monkeys, beetles and butterflies) to inanimate (cars, planes, chairs or cabinets). Distractors were visually scrambled, unrecognisable versions of these images. The images are square with the object centered and are displayed on a computer screen with a five degrees angle. The tasks were run on a PC using MatLab (2015a, The MathWorks) on a 120 Hz CRT-screen (ASUS) at a resolution of 1920 × 1080 pixels. Viewing distance was approximately 50 centimeters from the screen.
Procedure
The experiment consisted of two sessions of two hours. The first session participants completed the attentional blink task and on the second session the n-back task or vice versa. For the attentional blink each session consisted of 21 blocks (2 runs within each block), each run consisting of 40 trials. After reading the instructions participants started the trial by pressing the spacebar on the keyboard. A fixation cross appeared for 500 ms, after which a stream of 19 images are shown (2 unscrambled targets and 17 scrambled distractors). Each image is shown
for 20 ms followed by 80 ms of a blank screen, with a stimulus onset asynchrony (SOA: time between the first and second target) of 100, 200 or 700 ms (lag 1, 2 or 7 respectively) as is seen in figure 1. At the end of the stream a 6-item response menu was shown where participants had to decide which of the 6 images was the first target by using the corresponding keys (A, S, D, J, K, L) on the keyboard, see figure 2. After this, a new response menu appeared where the same was done for the second target. Between each block there was the possibility for an intermission.
Figure 1. Visual presentation of the operation of the attentional blink task. Images were shown
for 20 ms with an 80 ms blank screen between images. The time between targets differed between 100 ms, 200 ms or 700 ms (lag 1, 2 and 7 respectively).
Figure 2. Example of the response menu in the attentional blink task. Participants indicated
with the corresponding letters on the keyboard which image was shown at target 1. Another response menu with a new set of images was shown where the same was done for target 2.
An n-back task was used to measure working memory. The task consisted of 6 blocks (4 runs within each block), each run consisting of 100 trials (see figure 3). Each image is shown for 500 ms, with a variable SOA between 2.3-2.7 seconds. All participants went through the same condition, a 2-back task. Participants were asked to use buttons on the keyboard to indicate if the target image was the same as two images ago (Z for no, M for yes).
Figure 3. Visual presentation of the operation of the n-back task. Participants indicated if the
Results
From the 22 participants one participant had reaction times below 50 milliseconds and showed a significant drop in accuracy during testing. This participant was therefore excluded from the analysis. As can be seen in figure 4 (appendix) the n-back score and the AB-score on lag 7 were not normally distributed. A nonparametric Wilcoxon Signed Rank test was done to investigate the existence of a possible attentional blink. Results indicated that the accuracy at lag 7 was significantly higher than at lag 2 Z = -3.93, p < .001. This difference between lag 2 and lag 7 was expected and can be clearly seen in figure 5.
Figure 5. Mean proportion correct scores on the attentional blink task for lags 1, 2 and 7.
A Spearman correlation was run to determine the relationship between the n-back score and the AB-accuracy on lag 1. There was a weak correlation, which was significant: r(19) = .39, p = .04. This correlation is plotted in figure 6.
Figure 6. Distribution of scores of each participant on the n-back task.
Lastly, another Spearman correlation was done to assess the relation of the N-back score with the magnitude of the AB. Against expectations no such correlation was found, r(19) = .09, p = .356.
Discussion
The objective of this research was to determine the relation of the attentional blink with working memory. The manipulation to induce an attentional blink worked, there was a larger blink found in lag 2 than in lag 7. Second there was a positive relation between working memory and the accuracy on the first lag. Lastly, no link was found between the magnitude of the AB and working memory.
As expected, there was a positive link between working memory and accuracy on the first lag. Previous research on working memory and the attentional blink found that higher
memory loads led to lower attentional performance within the attentional blink period, but not beyond it (Akyürek, Hommel, & Jolicoeur, 2007). This would suggest a temporary lapse of attention that is limited to the early lags. Magnetoencephalography (MEG) - data indicates that the accuracy on the second target is related to the amount of attentional resources that are allocated to process the first target (Shapiro, Schmitz, Martens, Hommel, & Schnitzler, 2006). Although these results are all in accordance with each other, there is no justification that working memory is solely related to the attentional blink on the first lag. Here, timing was the important factor, performance on the second target was strongly affected by the temporal distance between the first and the second target (Akyürek, & Hommel, 2005).
As Colzato and colleagues (2007) stated: ‘a higher operation span may permit, or at least be associated with, longer integration windows.’ Lag-1 sparing could be the key explanation for the correlation between working memory and the attentional blink. Lag-1 sparing occurs when T1 and T2 are followed in close temporal proximity of each other. This sparing occurs when two targets are in the same temporal space when the attentional gate closes. They are transferred to working memory where it is possible to become aware of it. This could explain why only the first lag was found to be related to working memory, but not later lags. Sparing could even have an effect on the attentional blink. When three targets were in the same temporal space, reported targets were accompanied by significantly larger blinks than when they were missed (Dux, Wyble, Jolicœur, & Dell'Acqua, 2014). And even when sparing of both images is possible, they come into awareness at the loss of the temporal order of the targets (Hommel & Akyürek, 2005).
There were also unexpected results found. Since attention was presumed to benefit from working memory, it would logically follow that working memory would negatively correlate with the size of the attentional blink, although this was not found. A different aspect of memory could possibly explain these findings, but even multiple measures of working memory could
not find a relation with the magnitude of the AB (Martens & Johnson, 2009). Perhaps working memory is not the only underlying mechanism behind the attentional blink, excitement for or interest in the shown images induced an increased attentional blink (Beech et al., 2008). More research is needed to explain these results.
This experiment was meant as a pilot study to see if life-like images would yield similar results for accuracy as letters and numbers would in a classic attentional blink study. The images that were used could be stated as too easily recognized. This was actually the case, since a ceiling effect was found. Virtually all participants had very high scores on both the working memory task as the attentional blink task (see figure 6 and 7). If the images were too easily recognized, this would reduce the differences between the different lags and limit the effect of the attentional blink. Second, the fact that there was a learning effect found, could also be related to the images. The accuracy of the attentional blink task showed an upward trend as time progressed. This would indicate that previous trials had an effect on later trials. This study uses the same method as a previous attentional blink study (Marois & Chun, 2004). A similar masking method was also used, where images were scrambled and pixelated. An alternative explanation is therefore probable, the images themselves could have had an influence and perhaps be the cause of this learning effect. Lifelike images could be too easily recognised. Future research on this subject should control for these shortcomings which should increase the reliability of their results. The focus of future research should be on working memory. By determining it’s relation with lag-1 sparing it could be an important step in unravelling the mechanisms behind attention and ultimately our awareness.
Appendix
References
Akyürek, E. G., & Hommel, B. (2005). Target integration and the attentional blink. Acta Psychologica, 119(3), 305-314.
Akyürek, E. G., Hommel, B., & Jolicoeur, P. (2007). Direct evidence for a role of working memory in the attentional blink. Memory & Cognition, 35(4), 621-627.
Baddeley, A. D. (2002). Is working memory still working?. European psychologist, 7(2), 85.
Beech, A. R., Kalmus, E., Tipper, S. P., Baudouin, J. Y., Flak, V., & Humphreys, G. W. (2008). Children induce an enhanced attentional blink in child molesters. Psychological Assessment, 20(4), 397.
Broadbent, D. E., & Broadbent, M. H. P. (1987). From detection to identification: Response to multiple targets in rapid serial visual presentation. Perception and Psychophysics, 42, 105 113.
Carlson, S., Martinkauppi, S., Rämä, P., Salli, E., Korvenoja, A., & Aronen, H. J. (1998). Distribution of cortical activation during visuospatial n-back tasks as revealed by functional magnetic resonance imaging. Cerebral cortex (New York, NY: 1991), 8(8), 743-752.
Colzato, L. S., Spapé, M. M., Pannebakker, M. M., & Hommel, B. (2007). Working memory and the attentional blink: Blink size is predicted by individual differences in operation span.
Psychonomic Bulletin & Review, 14(6), 1051-1057.
DiCarlo, J. J., & Cox, D. D. (2007). Untangling invariant object recognition. Trends in cognitive sciences, 11(8), 333-341.
Dux, P. E., & Marois, R. (2009). The attentional blink: A review of data and theory. Attention, Perception, & Psychophysics, 71(8), 1683-1700.
Dux, P. E., Wyble, B., Jolicœur, P., & Dell'Acqua, R. (2014). On the costs of lag-1 sparing. Journal of Experimental Psychology: Human Perception and Performance, 40(1), 416.
Evans, K. K., & Treisman, A. (2005). Perception of objects in natural scenes: is it really attention free?. Journal of Experimental Psychology: Human Perception and Performance, 31(6), 1476.
Fahrenfort, J. J., van Leeuwen, J., Olivers, C. N., & Hogendoorn, H. (2017). Perceptual integration without conscious access. Proceedings of the National Academy of Sciences, 201617268.
Goodale, M. A., & Milner, A. D. (1992). Separate visual pathways for perception and action. Trends in neurosciences, 15(1), 20-25.
Gray, H. M., Ambady, N., Lowenthal, W. T., & Deldin, P. (2004). P300 as an index of attention to self-relevant stimuli. Journal of experimental social psychology, 40(2), 216-224.
Hommel, B., & Akyürek, E. G. (2005). Lag-1 sparing in the attentional blink: Benefits and costs of integrating two events into a single episode. The Quarterly Journal of Experimental
Psychology Section A, 58(8), 1415-1433.
Khosla, A., Raju, A. S., Torralba, A., & Oliva, A. (2015). Understanding and predicting image memorability at a large scale. In Proceedings of the IEEE International Conference on Computer Vision (pp. 2390-2398).
Kirchner, H., & Thorpe, S. J. (2006). Ultra-rapid object detection with saccadic eye movements: Visual processing speed revisited. Vision research, 46(11), 1762-1776.
Luck, S. J., Vogel, E. K., & Shapiro, K. L. (1996). Word meanings can be accessed but not reported during the attentional blink. Nature, 383(6601), 616.
Long, B., Yu, C. P., & Konkle, T. (2017). A mid-level organization of the ventral stream. bioRxiv, 213934.
Marois, R., Yi, D. J., & Chun, M. M. (2004). The neural fate of consciously perceived and missed events in the attentional blink. Neuron, 41(3), 465-472.
Martens, S., Munneke, J., Smid, H., & Johnson, A. (2006). Quick minds don't blink:
Electrophysiological correlates of individual differences in attentional selection. Journal of Cognitive Neuroscience, 18(9), 1423–1438.
Martens, S., & Johnson, A. (2009). Working memory capacity, intelligence, and the magnitude of the attentional blink revisited. Experimental Brain Research, 192(1), 43-52.
Martens, S., & Valchev, N. (2009). Individual differences in the attentional blink: The important role of irrelevant information. Experimental Psychology, 56(1), 18.
MATLAB and Statistics Toolbox Release 2015a, The MathWorks, Inc., Natick, Massachusetts, United States.
Shapiro, K., Schmitz, F., Martens, S., Hommel, B., & Schnitzler, A. (2006). Resource sharing in the attentional blink. Neuroreport, 17(2), 163-166.
Vogel, E. K., McCollough, A. W., & Machizawa, M. G. (2005). Neural measures reveal individual differences in controlling access to working memory. Nature, 438(7067), 500-503.
