Final Report Summary - LOADEMVS10 (The Role of Human Motivation in Visual Attention and Awareness under Load)
Perceptual load theory has been credited with solving the long-standing debate in Psychology about the locus of attentional selection. According to load theory, the degree to which unattended potentially distracting information can be ignored depends on the level of perceptual load involved in the task. In conditions of low perceptual load in the task spare capacity from the relevant processing “spills over” to the perception of an irrelevant distractor and attentional selection is late, post-perceptual. In conditions of high perceptual load that engage full perceptual capacity unattended distractors are not perceived. These effects are well established (e.g. Lavie, 2010 for review) however, load theory has not as yet been tested with motivationally salient stimuli. It is not clear whether tasks of high perceptual load allow people to ignore distractor information, which has high motivational salience. Information that has been associated through learning with a highly predictive outcome (e.g. monetary reward or loss) becomes motivationally salient. Such information should gain more attention than information that is less predictive of outcome. Thus, the main goal of this project was to explore the effects of high motivational value on visual selective attention and the associated neural correlates.
Specifically we have developed four lines of research. In one line, we investigated value-driven attentional capture under different perceptual load. To achieve this we conducted a series of experiments in which participants participated in a value-learning task in which six neutral faces were associated to different expected values (high-gain, high-loss, low-gain, low-loss, and no-outcome) through a betting game. Then these faces were used as distractors in a letter-search task. We found that high predictable faces captures attention and interferes with performance in a letter-search task compared to low predictable and neutral faces. We then presented these high predictable faces as distractors under low and high perceptual load task. We found that high-load (compared to low-load) significantly reduced distractor interference for high-loss faces, but had no effect on distractor interference produced by high-gain faces. These results suggest the extent of processing of unattended but highly predictable information depends on load and the valence of the outcome. To further test the effects of emotional valence in the next series of experiments we used different categories of positive and negative stimuli (such as International Affective Picture System (IAPS) pleasant and unpleasant images, happy, and angry faces) as distractors while performing a letter search task under low and high perceptual load. We found that high-load (compared to low-load) significantly reduced distractor interference for distractors with negative emotions, but had no effect on distractor interference produced by stimuli with positive emotions. These results thus converge with the results of the first study to show that the processing of unattended information of positive emotion does not depend on the task load, unlike negative or neutral information. In a third line of experiments, we investigated the interaction between the motivational salience and emotional valence of the information and expectancy: an important top down factor known to interact with attention, but not as yet known how it interacts with motivational salience, or emotional value. As in the second line three sets of distractor stimuli were used: faces associated with gain, loss, or neutral in an initial learning phase, positive or negative IAPS pictures, and happy or angry faces. Distractor expectancy was varied through the probability of distractor appearance. The results showed that positive emotional stimuli and those associated with gain captured attention and interfered with performance irrespective of the level of distractor expectancy. However, loss-associated and negatively valenced distractors interfere with performance only when it was presented unexpectedly. Taken together these results suggest that our visual system does not filter out the processing of positive information even in case of less attentional resources (under high load) or when highly predictable (presented with high probability) which renders their information value lower (Friston, 2005).
Load theory also discusses the role of working memory in selective attention. Working memory actively maintains the task processing priorities during task performance (for example between target and irrelevant distractors). It follows then that high working memory load (with task-unrelated material) will reduce the priority distinction between task-relevant and irrelevant stimuli and so working memory load has the opposite effect to perceptual load: increasing distractor processing. In a fourth line of experiments, we tested the effects of working memory load on motivationally salient distractors. We first investigated the effect of value-learned (motivationally salient) distractor faces on working memory performance. On each trial, two items (either faces or natural scenes) were presented for a period of 1s (the ‘memorandum’), then a task-irrelevant distractor face appeared for 2s, before a third face was shown (the ‘probe’), with ‘Respond’ displayed beneath it. Participants had to recognize whether the probe face was one of the two faces in the memorandum (by pressing one of two designated keys). We found that all value-coded face distractors (compared to neutral) interfere with working memory performance whether their content was congruent or incongruent with the memorandum. In addition, we also manipulated the working memory load of the memorandum and we found that under high-load condition, value-coded faces interfere working memory performance compared to neutral faces, but not under low load, where instead all distractors similarly influenced working memory performance.
Finally, we examined how early the processing advantage for motivationally salient stimuli is: does it extend to unconscious processing in the primary visual cortex (area V1). Participants learned the predictive value of faces with gain, loss, and neutral outcome in an initial learning phase. We then collected functional brain scans while these participants performed a face position discrimination task (high-gain, low-gain, high-loss, low-loss, or neutral faces were presented) using the continuous flash suppression (CFS) paradigm (Stewart et al., 2012). CFS was made up of dynamic noise patterns (frequency 10 Hz) generated by superimposition of shapes of random size and color at maximum contrast (Stewart et al., 2012). On each trial CFS was presented to one eye at full contrast while a value-coded face (either high-gain, high-loss, low-gain, low-loss, or neutral) was presented on a black background to the other eye and the contrast of the value-learned face was ramped up linearly by a 5% increment every 100 ms from 0% to 20% (subliminal presentation) during a trial. Our results revealed that V1, V2 and V3 responses in the retinal location of the invisible faces was significantly higher for high-predictable face (compared to low predictable faces). Higher response to the highly predictable faces compared to low-predictable faces was also found in orbital frontal gyrus (OFG) and the left insula. These results indicate that invisible faces associated with high motivational value can modulate activity in early visual cortex and higher brain areas like OFG and insula. This is an important contribution to value-learning research because modulation of preconscious neural activity in early visual cortex is in support of the theories of perceptual learning that posits that high motivational outcomes induces neural plasticity (Schultz, 2002) and alter neural mechanisms for analysis of incoming visual information.
Overall, this Marie Curie project has contributed to improve the knowledge of an important and ubiquitous role of human motivation and emotional value in visual attention, memory, and awareness under different levels of load. Our results are valuable to understand the way our brain deals with irrelevant yet motivationally salient information we are often exposed to while attempting to focus on our task-relevant processing.
Specifically we have developed four lines of research. In one line, we investigated value-driven attentional capture under different perceptual load. To achieve this we conducted a series of experiments in which participants participated in a value-learning task in which six neutral faces were associated to different expected values (high-gain, high-loss, low-gain, low-loss, and no-outcome) through a betting game. Then these faces were used as distractors in a letter-search task. We found that high predictable faces captures attention and interferes with performance in a letter-search task compared to low predictable and neutral faces. We then presented these high predictable faces as distractors under low and high perceptual load task. We found that high-load (compared to low-load) significantly reduced distractor interference for high-loss faces, but had no effect on distractor interference produced by high-gain faces. These results suggest the extent of processing of unattended but highly predictable information depends on load and the valence of the outcome. To further test the effects of emotional valence in the next series of experiments we used different categories of positive and negative stimuli (such as International Affective Picture System (IAPS) pleasant and unpleasant images, happy, and angry faces) as distractors while performing a letter search task under low and high perceptual load. We found that high-load (compared to low-load) significantly reduced distractor interference for distractors with negative emotions, but had no effect on distractor interference produced by stimuli with positive emotions. These results thus converge with the results of the first study to show that the processing of unattended information of positive emotion does not depend on the task load, unlike negative or neutral information. In a third line of experiments, we investigated the interaction between the motivational salience and emotional valence of the information and expectancy: an important top down factor known to interact with attention, but not as yet known how it interacts with motivational salience, or emotional value. As in the second line three sets of distractor stimuli were used: faces associated with gain, loss, or neutral in an initial learning phase, positive or negative IAPS pictures, and happy or angry faces. Distractor expectancy was varied through the probability of distractor appearance. The results showed that positive emotional stimuli and those associated with gain captured attention and interfered with performance irrespective of the level of distractor expectancy. However, loss-associated and negatively valenced distractors interfere with performance only when it was presented unexpectedly. Taken together these results suggest that our visual system does not filter out the processing of positive information even in case of less attentional resources (under high load) or when highly predictable (presented with high probability) which renders their information value lower (Friston, 2005).
Load theory also discusses the role of working memory in selective attention. Working memory actively maintains the task processing priorities during task performance (for example between target and irrelevant distractors). It follows then that high working memory load (with task-unrelated material) will reduce the priority distinction between task-relevant and irrelevant stimuli and so working memory load has the opposite effect to perceptual load: increasing distractor processing. In a fourth line of experiments, we tested the effects of working memory load on motivationally salient distractors. We first investigated the effect of value-learned (motivationally salient) distractor faces on working memory performance. On each trial, two items (either faces or natural scenes) were presented for a period of 1s (the ‘memorandum’), then a task-irrelevant distractor face appeared for 2s, before a third face was shown (the ‘probe’), with ‘Respond’ displayed beneath it. Participants had to recognize whether the probe face was one of the two faces in the memorandum (by pressing one of two designated keys). We found that all value-coded face distractors (compared to neutral) interfere with working memory performance whether their content was congruent or incongruent with the memorandum. In addition, we also manipulated the working memory load of the memorandum and we found that under high-load condition, value-coded faces interfere working memory performance compared to neutral faces, but not under low load, where instead all distractors similarly influenced working memory performance.
Finally, we examined how early the processing advantage for motivationally salient stimuli is: does it extend to unconscious processing in the primary visual cortex (area V1). Participants learned the predictive value of faces with gain, loss, and neutral outcome in an initial learning phase. We then collected functional brain scans while these participants performed a face position discrimination task (high-gain, low-gain, high-loss, low-loss, or neutral faces were presented) using the continuous flash suppression (CFS) paradigm (Stewart et al., 2012). CFS was made up of dynamic noise patterns (frequency 10 Hz) generated by superimposition of shapes of random size and color at maximum contrast (Stewart et al., 2012). On each trial CFS was presented to one eye at full contrast while a value-coded face (either high-gain, high-loss, low-gain, low-loss, or neutral) was presented on a black background to the other eye and the contrast of the value-learned face was ramped up linearly by a 5% increment every 100 ms from 0% to 20% (subliminal presentation) during a trial. Our results revealed that V1, V2 and V3 responses in the retinal location of the invisible faces was significantly higher for high-predictable face (compared to low predictable faces). Higher response to the highly predictable faces compared to low-predictable faces was also found in orbital frontal gyrus (OFG) and the left insula. These results indicate that invisible faces associated with high motivational value can modulate activity in early visual cortex and higher brain areas like OFG and insula. This is an important contribution to value-learning research because modulation of preconscious neural activity in early visual cortex is in support of the theories of perceptual learning that posits that high motivational outcomes induces neural plasticity (Schultz, 2002) and alter neural mechanisms for analysis of incoming visual information.
Overall, this Marie Curie project has contributed to improve the knowledge of an important and ubiquitous role of human motivation and emotional value in visual attention, memory, and awareness under different levels of load. Our results are valuable to understand the way our brain deals with irrelevant yet motivationally salient information we are often exposed to while attempting to focus on our task-relevant processing.