Evidence of higher level processing of neglected stimuli: comparison between residual
processing in hemineglect and unconscious processing observed in healthy subjects.
Klaudia Ambroziak, student number: 6249604 Supervisor: Johannes Fahrenfort Co-assesor: Annelinde Vandenbroucke
Abstract:
Neglect is a neurological condition usually caused by damage to the right parietal lobe, often following stroke. Patients with hemineglect tend to ignore objects or even physical stimulation coming from the visual field contralateral to the brain damage. Does neglect impair circuits that are involved in the perceptual analysis of visual input, or is it an attentional deficit that leaves perceptual processing of neglected objects intact? It has often been assumed that in neglect, the stimuli in the visual field contralateral to the brain damage are not elaborately and that processing is confined to very early stages. This account explains neglect as a problem of early selection of information. However, a vast body of literature shows that there is substantial perceptual processing in neglect: emotional and social processing, semantic processing, and processing of visual illusions. This review discusses different types of processing and addresses the question of its limits. Is this processing restricted to the lower levels or does it involve more elaborate processing? The processing of stimuli presented to the neglected side is compared with unconscious processing observed in healthy subjects, e.g. during masked priming. The evidence suggests that processing of neglected stimuli is not abolished during early stages, but involves higher level, more elaborate processing.
1. Introduction 3 1.1. Neglect and extinction.
1.2. Unconscious processing in healthy subjects.
2. Emotional processing. 5
2.1. Emotional processing of unconscious stimuli in healthy subjects. 2.2. Emotional processing in neglect.
2.3. Conclusions.
3. Social processing. 8
3.1. Unconscious social processing in healthy subjects. 3.2. Social processing in neglect patients.
3.3. Conclusions
4. Semantic and categorical processing. 10
4.1. Unconscious semantic processing in healthy subjects. 4.2. Semantic processing in neglect patients.
4.3. Conclusions.
5. Processing of visual illusions. 14
5.1. Unconscious processing of illusions by healthy subjects. 5.2. Processing of visual illusions in neglect patients.
5.3. Conclusions.
6. Neural correlates. 17
6.1. Neural correlates of unconscious processing. 6.2. Neural correlates of neglect.
6.3. Discussion.
1. Introduction.
1.1. Neglect and extinction
Neglect, also called hemineglect or parietal neglect, is a neurological condition usually caused by damage to the right parietal lobe often following stroke. Patients with hemineglect tend to ignore objects or even physical stimulation coming from the visual field contralateral to the brain damage. They do not react to and do not search for these stimuli. In a real life situation, the deficit can reach the point where patients make up or shave only one half of their face, eat from one side of a plate and so on. In clinical or experimental testing, when patients with neglect are presented with a picture and asked to draw it, as a result they copy only one half of it (see Fig. 1).
Figure 1. A model drawing (at the top) and its copy made by a patient with neglect. (from Odgen, 2012) The degree of impairment can vary among patients. However, if their attention is directed to the neglected field, they are usually able to consciously perceive stimulation from this side. For example, patients can notice the fence in the picture (Fig. 1) after it is pointed out directly. When they are asked why they did not copy the fence in their drawings, they may come up with different and sometimes remarkable explanations, such as "the fence will probably blow down in the next wind, so there is no point in drawing it” (Odgen, 2012)
Similar to neglect is a condition known as extinction. Patients with extinction ignore stimuli on the side contralateral to the lesion if there is a simultaneous stimulation on the ipsilateral side. In most cases they are able to normally detect stimuli on the contralateral side if they are presented in separation. It is commonly thought that extinction is a milder form of neglect, and these two syndromes are often studied together. Some researchers argue that neglect and extinction are different deficits and should be studied separately (de Haan, Karnath and Driver, 2012). This review, however, adopts the traditional approach.
Both in neglect and extinction, the visual cortex is intact and information can still be processed to some extent. What is the nature of this processing? Does neglect impair circuits that are involved in the perceptual analysis of visual input, or is it an attentional deficit that leaves perceptual processing
of neglected objects intact? It is a well-established fact that visual information is processed in a hierarchical way. During the early stages low-level features are processed, whereas the later stages are characterized by high-level processing. It has often been assumed that in neglect, the stimuli in the visual field contralateral to the brain damage are not elaborately processed and that processing is confined to very early stages. This account explains neglect as a problem of early selection of information. However, a vast body of literature shows that there is substantial residual processing in neglect. In a classic study by Marshall and Halligan (1988), patient P.S. was presented simultaneously with two pictures of a house, one of which the house had flames coming from the left side. She judged that the drawings were identical. However, when asked to select which house she would prefer to live in, she reliably chose the house that was not burning.
The question addressed here is about the limits of the residual processing of neglected stimuli.Is this processing restricted to the lower levels or does it involve more elaborate processing? I will compare the processing of stimuli presented to the neglected side with unconscious processing observed in healthy subjects, e.g. during masked priming. The evidence presented in this paper for both neglect patients and healthy subjects is divided into four main categories: emotional processing, social processing, semantic and categorical processing, and processing of illusory figures. However, this is not to say that the experiments presented in each section can only be explained with reference to a single type of processing. In other words, the burning houses experiment can be an example of both emotional and semantic processing.
1.2. Unconscious processing in healthy subjects.
It has traditionally been assumed that unconscious processing operates on the lower levels. It was characterized as automatic, pre-attentive, or operating on simple associations, e.g. motor preparation. A main characteristic of an automatic reaction is being spontaneous and rapid, as well as occurring without attention or conscious awareness. Unconscious processing can occur in situations when stimuli are masked from awareness, and thus unconscious, even if spatial attention is directed towards the stimuli. Conscious processing, on the other hand, is thought to be dynamic, flexible, adaptive, context specific, elaborated and goal-directed. However, new evidence suggests that unconscious processing may have characteristics that were traditionally attributed to conscious processing. The type of information that can be extracted during unconscious processing is now widely debated.
There are several methods that can be used to make stimuli invisible, including backward masking, continuous flash suppression (CFS) and interocular suppression. In backward masking, one visual stimulus (mask) is shown immediately after another visual stimulus (target) which prevents conscious perception of the first stimulus. In CFS, one eye is presented with a static stimulus (target) while the other eye is presented with a series of flashing stimuli (distractors). CFS works similarly to interocular suppression in which when one eye is shown a constantly moving visual pattern to suppress a stationary image presented to the other eye.
2. Emotional processing.
In the study by Marshall and Halligan (1988), described in the previous section, patient P.S. chose the house that was not burning, when presented with two pictures of a house, one in which the left side was on fire, even though she consciously claimed that the pictures were identical. Since the preference for the non-burning house was persistent over time, it can be claimed that patient P.S. processed the left side of both houses. Fire in the burning house served as a threatening stimulus and was able to influence patient’s choice. This case shows an example of how fear related stimuli can be processed by a patient with neglect. In this section I will look further intospecial aspects of emotional processing. I will compare processing of stimuli presented to the neglected field with processing of masked stimuli by healthy subjects.
2.2. Emotional processing of unconscious stimuli in healthy subjects.
Research on affective priming shows how unconsciously processed stimuli can affect subjects’ behavior. In their study, Marshall and Zajonc (1993) used faces with emotional expressions (happy and angry) as primes and Chinese ideographs as primed targets. The ideographs were selected as being neutral and novel. Subjects were told they would be presented with a set of Chinese characters and that they had to indicate on a scale from 1 to 5 how much they like each character. Before each target appeared, the subjects were presented with an emotional face for 4 ms. The study revealed that subjects preferred ideographs that were preceded by positive facial expressions. These findings show that emotional stimuli presented for only 4 ms can influence an individual's perceptions. As a consequence, discrimination between positive and negative expression can be made outside of conscious awareness.
Recent research (Anderson, Siegel, White and Barrett, 2012) shows similar results when emotional primes (e.g. happy or angry faces) were made invisible by continuous flash suppression (CFS). Participants had to rate neutral faces as pleasant or unpleasant, as more or less trustworthy, likable, competent and attractive. Results demonstrated that information from unseen emotional faces influences judgments of neutral faces. In particular, participants judged neutral faces more positively when paired with invisible smiling faces, and more negatively when paired with unseen angry faces. More evidence comes from studying facial muscles’ reactions to emotional expression. When people are exposed to emotional faces (e.g. smiling or angry), they spontaneously react with distinct facial muscles movements, as if they were to mimic expressions of the stimuli (Dimberg, 1982). Dimberg, Thunberg and Elmehed (2000) investigated whether similar facial reactions can still be evoked when people are exposed to happy and angry faces unconsciously. In their experiment the backward-masking technique was used to prevent subjects from consciously perceiving target faces with emotional expressions. Participants were divided into 3 groups. The first group was exposed to happy faces, the second group to angry faces, and the third group to neutral faces. In all groups, every target face was presented only for 30 ms and was immediately followed and masked by a neutral face. Facial reactions were measured by EMG electrodes. Subjects were not told about the real purpose of the electrodes, and they were not aware that their facial activity was measured. Results showed
that the subjects reacted with distinct facial muscle reactions that corresponded to the happy and angry stimulus faces. Thus, both positive and negative emotional reactions can be evoked without awareness of the stimuli.
2.1. Emotional processing in neglect.
A considerable body of evidence shows how emotional and fear related stimuli can be processed in patients with hemineglect. In some cases, like in the experiment by Marshall and Halligan (1988), emotional stimuli presented in the neglected field can influence patients’ behavior without entering awareness. In other cases, emotional stimuli in the neglected field can be processed to the level where patients are able to report about them. The same patients usually miss neutral visual stimuli on the contralesional side. Vuilleumier and Schwartz (2001) studied patients with unilateral neglect and visual extinction to see whether threatening stimuli can overcome attentional deficit in these patients. In the experiment, pictures of spiders were used as fear-related stimuli, whereas pictures of flowers and rings were used for comparison (see Fig.2.). Stimuli were presented in either right, left, or both visual fields. Patients had to locate and identify the stimuli on each trial by verbally naming their type and position e.g. “flower on the right and nothing on the left”. Critically, pictures of spiders and flowers were designed in such a way that they shared exactly the same visual features (see Fig.2.). This was done by shifting the legs of the spiders to make them resemble flowers' petals. Thus, a difference in perception for these stimuli could not be explained by differences in their low-level visual attributes, e.g. the degree of contrast or brightness.
Fig. 2: Stimuli used by Vuilleumier and Schwartz (2001).
Both patients showed a marked extinction of contralesional stimuli on bilateral trials. However, results showed that the patients detected fear-related pictures of spiders on the contralateral, left side much more often than neutral pictures of flowers. This shows that the rate of extinction can be modulated by the nature of the contralesional stimulus. Importantly, since the pictures of spiders and flowers shared the same low-level attributes, these results suggest that processing of neglected stimuli goes up to high-level visual representations. The authors concluded that, while mechanisms of spatial attention are impaired after parietal damage in neglect patients, intact visual pathways to
the ventral temporal lobe and amygdala may still mediate distinct mechanisms of emotional attention.
A well-studied type of emotional stimulus is facial expression. Lucas and Vuilleumier (2008) investigated the effects of neutral and emotional facial expressions on visual search in neglect patients. Participants without neglect were tested for comparison. Both groups performed a simple visual search task in which they had to report the gender of a target face. The target identity was unique among an array of 8 distracters, which were neutral faces with a different identity. There were three possible ways in which the target face could differ from the distracters. In the first condition, the target face differed from the distracters by identity alone, and had a neutral expression. In the second condition, the target face differed from the distracters by identity and color (red shade). In the third condition, the target face differed from the distracters by identity, and had an emotional expression: fearful or happy. Results showed that although neglect patients were slower in detecting target faces on the neglected side of the display, they were still influenced by the different cueing conditions. Reaction times for both color-cued and emotionally-cued faces were faster in comparison to the neutral face targets. Healthy subjects showed the same cueing pattern, but they were much faster in detecting targets on their left side. Therefore, despite attentional deficit in their left hemifield, neglect patients may nevertheless processed not only low-level color cues, but also emotional cues.
The studies described above stress the role of amygdala in emotional processing. A large body of literature shows that activity in the amygdala differs when people are exposed to different facial stimuli, and that damage to the amygdala impairs the recognition of emotional expressions (Morris, Ohman and Dolan, 1998; Adolphs, Tranel, Damasio, & Damasio, 1994). Parietal damage in neglect patients impairs mechanisms of spatial attention, but processing of emotional stimuli may still be mediated by intact visual pathways to the ventral temporal lobe and amygdala. Vuilleumier, Armony, Clarke, Husain, Driver and Dolan (2002) studied patients with neglect to compare the neural response in amygdala elicited by emotional faces with and without awareness. Comparison between perceived and neglected fearful faces showed no difference in amygdala activation. These results demonstrate that the amygdala can be activated by emotional stimuli even without awareness.
2.3. Conclusions
Both studies of neglect patients and affective priming research indicate that emotional stimuli can be processed outside of awareness. In patients with neglect, emotional stimuli may in some cases overcome the attentional deficit, in contrast to neutral stimuli (Vuilleumier and Schwartz, 2001). Affective priming research show that unconscious emotional stimuli can influence behavior (Anderson et al., 2012; Marshall and Zajonc, 1993) and elicit corresponding facial reaction in healthy subjects without entering awareness (Dimberg et al., 2000).
This greater salience of emotional stimuli can be explained from an evolutionary point of view. Fast processing of threat-related information is important for survival and it is the foundation of adaptive behavior. An angry face or spider-like shape may be a sign of serious danger. Thus, fear related
stimuli have a privilege in capturing attention, which cause faster detection even outside of attention. In this sense, emotional processing has to be automatic. On a neural level this fast processing of emotional stimuli and its greater attentional saliency can be explained by an emotional modulation from amygdala, independent of attentional influences from the parietal network damaged in neglect (Vuilleumier et al. 2002).
Taking an evolutionary perspective would suggests that unconscious emotional processing is mainly automatic. Furthermore, most affective priming studies use simple stimuli or faces. Processing of faces may have a special place in visual system e.g. fusiform face area is thought to be, as suggested by its name, specialized in facial recognition. Thus, it can be argued that emotional processing is rapid and automatic. However, the results presented in this section seems to support the claim that processing of neglected stimuli is not resolved at a lower level. Faces with emotional expressions of fear may be better detected than neutral faces with the same low-level features (Lucas and Vuilleumier, 2008). Moreover, fear-related stimuli such as spiders may, contrary to neutral flower stimuli, overcome the attentional deficit, even if spider and flower stimuli share the same low-level characteristics (Vuilleumier and Schwartz, 2001). Therefore, it can be claimed that pre-attentive processing of neglected stimuli can result in higher-level representations before stimuli are categorized as fear-related or safe.
Some questions about the nature of emotional processing arise. How thorough can the processing of affective stimuli be? Is it the case that emotional stimuli are elaborated in detail, or is it only its emotional content that gets processed? From the studies presented above it may appear that the former is true about neglect, and the latter about unconscious priming. However, it is possible that during unconscious priming, emotional content of stimuli may automatically capture attention outside of awareness and then facilitate further processing which is carried on to the later stages. The study by Jiang et al. (2006), presented in the next section, shows that healthy subjects can unconsciously discriminate between gender if stimuli are sexually arousing, whereas findings for the neutral faces are mixed.
3. Social processing.
Studies presented in the previous section showed that emotions can be interpreted from facial expressions without conscious awareness of the face. The question addressed in this section is whether other information such as gender or race could also become available without awareness of the face.
3.1. Unconscious social processing in healthy subjects.
The research of Anderson et al. (in press) described in a previous section showed that when faces were made invisible by continuous flash suppression (CFS), facial expressions could influence participants’ judgments. Emotional information may be unique in this respect. Amihai, Deouell and Bentin (2011) investigated whether CFS can abolish processing of detailed information necessary for categorization of the face as male or female, Asian or European. Previous studies (Webster,
Kaping, Mizokami and Duhamel, 2004) showed that if a prime face was visible participants were biased away from the gender and the race of the prime when they estimated the gender and the race of the ambiguous target face. If the prime face was clearly male, the ambiguous target face tended to be perceived as female (the gender aftereffect). Amihai et al. (2011) tested whether similar bias could be observed when prime faces were made invisible by (CFS). Participants were presented with unfamiliar ambiguous faces and were asked to classify their race or gender. Ambiguous faces were preceded by invisible faces with clearly determined gender and race. The occurrence of bias would indicate that information about gender and race could be processed outside of conscious awareness. The results showed, however, that when prime faces were invisible, participants were no longer biased in their judgments. Nevertheless, different results were found by Shin, Stolte and Chong (2009). In this study subjects also had to judge ambiguous faces as male or female. The ambiguous stimuli were obtained by morphing male and female faces. The male–female ratio varied among stimuli. The points of subjective equality (PSE) were measured and compared before and after adaptation to unseen face stimuli. Female faces were used as adaptors and rendered invisible by interocular suppression. During the adaptation phase, participants detected contrast decrements in one of the two suppressors. The strength of adaptation was indicated by as the ratio between the pre- and post-adaptation PSE. This change in PSE would be caused by the gender aftereffect. The occurrence of the gender aftereffect would indicate that information about gender was processed outside of conscious awareness. The results showed that the strength of adaptation to the high-level features of perceptually invisible stimuli (i.e. their gender) was increased by enhancing spatial attention to this invisible target.
A question that arises is whether additional emotional stimulation can influence processing of the high-level features of invisible stimuli. Amihai at al. (2011) used moderately attractive neutral faces as invisible primes and found no evidence for unconscious processing of gender. However, using arousing pictures of sexually attractive people can show different results. Jiang, Costello, Fang, Huang and He (2006) tested both heterosexual and homosexual men and women to investigate whether invisible erotic images can guide spatial attention. Participants had to indicate the orientation of the Gabor patch briefly, and randomly presented to either left or right side of fixation point . The Gabor probe was preceded by a cue display with an erotic picture of either a nude male or female shown on one side of the display, and a scrambled control image on the other side. Interocular suppression was used to make the cue display invisible. The attentional effect was measured by the difference in accuracy between the condition when the Gabor probe was presented on the side of the erotic image, and the condition when the Gabor probe was presented on the side of the scrambled control. Results showed that participants performed better in the first condition. For all groups, accuracy was higher when erotic pictures presented people of the preferred gender. Furthermore, heterosexual male participants performed significantly worse when erotic pictures presented nude males. Thus, invisible erotic stimuli can either attract or repel attention depending on gender and the sexual orientation of participants. These results suggest that information about gender can be process even if stimuli are not consciously perceived. Thus, it can be argued that during unconscious priming, emotional content of stimuli can facilitate further processing which is carried on to the later stages.
3.2. Social processing in neglect patients.
Evidence shows that healthy subjects can discriminate between gender if stimuli are sexually arousing, but for neutral faces the findings are mixed. Fewexperimental studies of neglect patients address this question directly. The experiment by Lucas and Vuilleumier (2008), presented in the previous section, suggests that the gender of faces is processed pre-attentively in hemineglect patients. In this study, participants performed a visual search task in which they had to report the gender of a target face with the identity unique amongst an array of 8 distracters. The results showed that reaction times were slower in conditions when the target and distracters had the same gender, than in conditions when the target and distracters had different gender. Furthermore, slowing caused by gender similarity was greater on the contralesional side as compared with the ipsilesional side. This effect cannot be explained by low-level characteristics. Greater pictorial similarity between different genders than different identities indicate that low-level differences between different genders were smaller then between faces of different identity but same gender. Thus, if discrimination between faces was resolved at a lower level, different identity should attract attention regardless of their gender.
3.3. Conclusions.
The evidence shows that during unconscious processing healthy subjects can discriminate between gender if stimuli are sexually arousing but for neutral faces the findings are mixed. It can be argued that when stimuli are highly arousing, the emotional aspect plays crucial role. Moderately attractive neutral faces had no priming effect in a study by Amihai et al. (2011). On the other hand, Shin et al. (2009) showed that the strength of adaptation to the features of perceptually invisible stimuli i.e. gender, is increased by enhancing spatial attention. Whereas discriminating between genders of faces may require awareness or attention directed towards stimuli, sexually arousing stimuli can direct the distribution of spatial attention outside of awareness.
Some evidence suggests that gender can be pre-attentively interpreted from faces presented to the contralateral side by neglect patients. However, very few experimental studies address this question directly. Future studies should investigate the issue of the processing of social cues by neglect patients more in depth. In particular, the question of whether gender can be correctly interpreted from a neglected face stimulus without attracting attention should be addressed.
4. Semantic and categorical processing.
The next type of processing involves semantic and categorical information. Semantics investigate the relation between symbols and their meanings. These symbols can be letters and words, digits and numbers, or signs. Thus, semantic processing depends upon the ability to operate on abstract representations and categories, e.g. on manipulating numbers. It is commonly thought to require complex, higher-level categorization. Some of the studies already presented in previous sections may also serve as examples of semantic processing in patients with neglect. The famous burning houses experiment (Marshall and Halligan, 1988) was recalled asan illustration of emotional processing of
fear related stimuli. However, it can be also seen as an example of semantic processing. Fire in the pictures presented to the patient very vaguely resembles the real threat. To discriminate between the safe and the burning house, patients had to construct some type of higher level semantic representations of neglected parts of the images. Likewise, the experiment by Vuilleumier and Schwartz (2001) comparing the efficiency of spider and flower images in overcoming attention deficit, suggests that processing of neglected stimuli goes up to high-level visual representations. The pictures of spiders and flowers used in this study, shared the same low-level attributes. Therefore, participants could not categorize them using low-level characteristics. In this section I will present and discuss more evidence for semantic processing in neglect, and compare it with examples of unconscious semantic processing.
4.1. Unconscious semantic processing in healthy subjects.
There is an ongoing debate whether unconscious semantic processing is indeed possible. Astudy by Dehaene, Naccache, Le Clec’H, Koechlin, Mueller, Dehaene-Lambertz, Moortele and Bihan (1998) is an example of a masked priming experiment which suggests that semantic processes can occur without awareness. Twelve subjects performed a number comparison task during brain imaging (EEG and fMRI). The subjects were shown target numbers between 1 and 9 (excluding 5), and requested to indicate by a button press whether the target was larger than 5 or smaller than 5. The targets were randomly selected among the set of four numbers 1, 4, 6, 9, presented equally often in Arabic notation or in verbal notation. Each target number was preceded by a number primesurrounded by masks that made it invisible. The set of primes was the same as the set of targets. Behavioral results revealed that subjects were faster when the prime and target numbers fell on the same side of 5, and therefore required the same motor response. This priming effect was significant even when the notation of the target and the primes were different, i.e. when one number was presented in Arabic notation and the other in verbal notation. Furthermore, measures of motor activity showed that subjects were preparing a covert motor response appropriate to the prime before giving the overt motor response appropriate to the target. The authors concluded that information from the invisible prime was unconsciously processed through a series of stages that included the semantic categorization of the prime as larger or smaller than 5, and then proceeded all the way to a motor level. However, a strong objection against this interpretation was raised. It can be argued that the results may also be explained non-semantically by the direct motor specification hypothesis (DMS; see Neumann and Klotz, 1994). According to the DMS hypothesis, when subjects consciously process stimuli, they establish stimulus–response associations that can later be elicited during unconscious processing of masked stimuli. This type of association can be activated automatically and therefore does not demonstrate unconscious access to semantic representations. In many priming paradigms, stimulus– response learning is possible because a small set of stimuli is used repeatedly. Thus, most masked priming results can be explained by low-level direct stimulus–response associations.
Nevertheless, evidence shows that genuine semantic processing of unconscious information may be possible. Naccache and Dehaene (2001) argue that unconscious semantic priming can be extended to novel stimuli, and that unconscious access to semantic information is possible. In this experiment they used a similar task as Dehaene et al. (1998). As in the previous experiment, subjects also had to
indicate by button press whether the target was larger than 5 or smaller than 5. A target could again be the number 1, 4, 6, or 9 in either Arabic or verbal notation. However, this time a prime could be either the neutral symbol $ or the number 1, 2, 3, 4, 6, 7, 8, or 9 in Arabic or verbal notation. Thus, the numbers 1, 4, 6, and 9, which were used as targets, formed the old set of primes, while the numbers 2, 3, 7, and 8, which were never used as targets, formed the new set primes. Their study showed that unconscious semantic priming occurs even for prime stimuli that are never presented as target stimuli, and for which no stimulus response learning could conceivably occur.
Furthermore, in a recent study Wokke, van Gaal, Scholte, Ridderinkhof and Lamme (2011) demonstrated that the same unconscious stimulus can have a different effect on behavior and brain activity depending on the context in which it is presented. The subjects performed a Go/No-Go task, instructed to respond as fast as possible to a Go target by pressing a button. A target could be a diamond or a square shape. At the beginning of each trial a cue indicated which of both stimuli (square or diamond) functioned as a No-Go target in the upcoming trial. In every trial there was an unconscious (masked) prime preceding a target. The same prime stimulus was associated with a No-Go response on one trial, but with a Go response on the next (depending on the instruction cue). This manipulation hindered the formation of stable associations between primes and targets. Hence unconscious stimuli were prevented from activating the actions automatically. Results showed that the reaction times were significantly slower to Go targets preceded by a No-Go prime, compared to the reaction times on Go targets preceded by a Go prime. Participants inhibited their responses more often when a No-Go target was preceded by a No-Go prime than when it was preceded by a Go prime, indicating that unconscious primes affected inhibitory performance on subsequent No-Go targets. By eliminating stimulus-response associations established through conscious learning, this study showed that unconscious information can be processed in a flexible and adaptive manner. This suggests that unconscious processing may have characteristics that were traditionally attributed to conscious processing.
4.2. Semantic processing in neglect patients.
A study by Sackur, Naccache, Pradat-Diehl, Azouvi, Mazevet, Katz, Cohen and Dehaene (2008) uses a paradigm similar to that of Dehaene et al. (1998), and demonstrates semantic processing of neglected numbers. Four patients with unilateral left spatial neglect performed a number comparison task. Each target number was preceded by a lateralized number prime, either in the intact or neglected hemifield (HF). A screen with a numerical prime on one side and a distractor stimulus on the other appeared for 200 ms and was followed by the target number which stayed on the screen until a response was made. The prime and target were randomly selected among the set of four numbers 1, 4, 6, 9, presented equally often in Arabic notation or in verbal notation . The prime could be either congruent (same category as target) or incongruent. The distractor was a meaningless pattern obtained from scrambling the letters of the word ‘‘QUATRE’’. Patients were instructed to indicate whether the target number was smaller or larger than five. As expected, congruent primes facilitated the response: results showed that congruent trials were faster than incongruent trials. Importantly, there was no significant difference in reaction times between primes presented to the left (neglected) field or to the right (intact) field. This absence of interaction
between the priming effect and the prime hemifield suggests that primes were processed in a very similar way in both hemifields. Even if the number was presented in the neglected field, participants were able to unconsciously recognize, for example, that ONE and 1 represent the same number, and that ONE and 4 fall on the same side of 5. This type of reasoninggoes beyond low-level information. Similarly to the study by Dehaene et al. (1998), this experiment was limited by the fact that the set of left primes was exactly the same as the set of targets. Therefore, the results can be reinterpreted non-semantically within the DMS framework: semantic processing during conscious trials may serve as a basis for the setting up of lower level sensory motor associations that were active during unconscious trials, producing the unconscious congruity effect. However, this effect was already present very early during the experiment (first 32 trials), before stimulus–response association could be established. This indicates that the priming effect does not depend on extensive learning, as would be expected if the effect resulted from DMS.
Berti and Rizzolatti (1992) show that patients with neglect are able to process stimuli presented to the neglected field to a categorical level of representation even when they deny stimulus presence in the affected field. In this study seven patients were tested with a semantic categorization task where the influence of priming stimuli, presented to the left (neglected) visual field, was measured by manual reaction times to the target stimuli, presented to the right (non-neglected) visual field. Stimuli were line drawings of objects belonging to one of two categories: animals or fruits. Prime stimuli were always green and target stimuli were red. Subjects were instructed to respond only to the red stimuli, and indicate their category by pressing one of two keys. Three experimental conditions were introduced: "Highly congruent", "Congruent" and "Noncongruent". In the "Highly congruent" condition the target and prime stimuli belonged to the same category and were physically identical, e.g. two apples. In the "Congruent" condition the stimuli represented two different exemplars of the same category, e.g. an apple and grapes. In the "Noncongruent" condition the stimuli represented one exemplar from each of the two categories of stimuli, e.g. an apple and a mouse. Five out of seven patients never reported the stimulus presented to the left of the fixation point, claiming that they were seeing only one object and not two. The other two subjects could sometimes report the stimuli on the contralateral side, so they were excluded from further analyses. The results showed that the reaction times for "Noncongruent" were slower than RTs in the other two conditions. No difference was found between the “Highly Congruent” and “Congruent” conditions. Thus, responses were facilitated not only in the “Highly congruent” condition, but also in the “Congruent” one. If neglect patients did not process priming stimuli presented in the neglected field, this effect would not be found. Furthermore, if processing in the neglected field was limited to low-level visual analysis, the facilitation would only be found in a condition in which there is a complete congruence (both physical and semantic identity) between the prime and target.
4.3. Conclusions.
Evidence suggests that semantic processing is possible both in neglect and during unconscious visual stimulation in healthy participants. In both cases, the results showed that participants unconsciously recognize, for instance, that ONE and 1 represent the same number, and that ONE and 4 fall on the same side of 5 (Dehaene et al., 1998, Sackur et al. 2008). This indicates that the priming effect
does not depend on extensive learning, as would be expected if the effect resulted from DMS. The presence of a congruency effect with minimal learning and its independence from notational effects, on the other hand, is suggestive of semantic-level cognitive processing rather than the learning of a direct motor chain. Furthermore, the research of Naccache and Dehaene (2001) showed that in healthy subjects, the results of unconscious semantic priming cannot be explained by the contribution of a DMS. In this study novel primes were used, which never appeared as targets in the experiments, and thus could not be easily associated with a motor response. In the future, a similar manipulation should be used with neglect patients to strengthen the evidence for semantic processing of neglected stimuli.
Apart from that, it was suggested that unconscious processing may have characteristics that were traditionally attributed to conscious processing, e.g. flexibility and context-specificity (Wokke et al. 2011). While, patients with neglect are able to process stimuli presented to the neglected field to a categorical level of representation (Berti and Rizzolatti, 1992).
5. Processing of visual illusions
The next type of evidence comes from studies on perceiving visual illusions in neglect patients. Visual illusions are especially interesting for researchers studying conscious perception, since in illusions actual physical stimulation differs from what is consciously perceived. The evidence suggests that some of the simple illusions can be explained by low-level processing, while other, more complex ones require high level processes. Thus, the ability to perceive certain illusions may serve as a measure of conscious awareness. An example of the complex illusion may be the Kanizsa subjective contours illusion where subjects perceive illusory contours of figures, which are not really there, guided by inducers - usually circles with missing wedges. Although contours are not delimitated by luminance or color change, illusory figures often appear to be brighter than the background.
5.1. Unconscious processing of illusions by healthy subjects.
A recent study by Harris, Schwarzkopf, Song, Bahrami and Rees (2011) showed that awareness is necessary to perceive the Kanizsa illusion but not the simultaneous brightness contrast (SBC). Harris at al. used continuous flash suppression (CFS) with a Kanizsa triangle presented to one eye, and a rapidly changing mask consisting of randomly positioned colored rectangles and ellipses presented to the other eye. In one condition, a single CFS mask covered the entire stimulus, while in the second, i.e. the selective-mask condition, four circular CFS masks covered only the inducers in the illusion stimulus. The results showed that when inducers were suppressed from awareness using CFS, subjects did not experience illusory contours. Rather, their performance dropped to chance level compared to control trials with non-suppressed stimuli. In a separate task Harris at al. used the simultaneous brightness contrast: an illusion of a white circle seeming brighter when it is presented on a black background than when it is presented on a (dark) grey background. Again, one eye was presented with a CFS mask that allowed for controlled suppression of the illusion, which was
presented to the other eye. The results showed that a simultaneous brightness contrast persisted even when the CFS was used to make the surrounding context invisible.
This study shows that CFS does not abolish low-level processing of simple illusions (like a brightness contrast illusion), but is sufficient to suppress the Kanizsa illusion. This suggests that conscious processing of spatial context is necessary for Kanizsa illusion to occur and that high-level inferential mechanisms are involved. Thus, it can be argued that perception of the Kanizsa illusion depends on high-level inferential processes, and implies awareness of the inducing elements, while simple illusions like SBC are driven by low-level stimulation.
5.2. Processing of visual illusions in neglect patients.
A number of studies show that perception of illusory contours (Kanizsa illusion) is preserved in patients with hemineglect and extinction (Mattingley, Davis and Driver, 1997; Vuilleumier and Landis, 1998; Vuilleumier, Valenza and Landis, 2001; Conci, Böbel, Matthias, Keller, Müller and Finke, 2009). In a study by Vuilleumier, Valenza and Landis (2001) twelve patients were tested in two separate tasks for implicit and explicit detection of Kanizsa illusory contours. All subjects had right hemisphere damage and left neglect. Stimuli (see figure) were printed in black on a white sheet of paper and presented on a table desk. In the first task patients were not explicitly required to attend to lateral elements of the stimuli, but asked to judge the length of the figures and mark their midpoint by drawing a small dot at their centre. Bisection judgments for Kanizsa-type illusory bars and rectangles (Fig. 3a and b) were compared to bisection judgments for two types of control stimuli. The first type included horizontal lines (Fig. 3c), bars and rectangles delimited by real contours (figure d). The other type consisted of gap figures delimited by two vertical lines having a comparable height and horizontal extent, but producing no subjective grouping and filling-in (Fig. 3e).
The results of bisection judgments for illusory Kanizsa figures were similar to those of the physically connected stimuli (lines and real contours figures), and showed significant rightward deviation. On the other hand, bisection judgments for the gap stimuli significantly differed from the three other stimuli condition: there was significantly less rightward deviation for the gap stimuli compared to Kanizsa or real contours figures. This suggests that the spatially unconnected features in Kanizsa figures were effectively grouped together and treated as a single object for midpoint judgments, unlike the non-illusory gap figures. In the second task patients were shown the same Kanizsa illusory figures, but this time they had to overtly attend to lateral elements of the stimuli and decide whether inducers on the right and left side were the same (figure a) or different (figure b). Patients could accurately report all differences in inducers on the right (ipsilateral) side of Kanizsa figures but they usually failed to detect differences in inducers on the left (contralateral) side of the figures (mean 70% missed). Together, these findings show that perceiving illusory contours can occur pre-attentively and influence subjects’ performance independently from the ability to detect contralateral inducers explicitly.
5.3. Conclusion.
The evidence shows that perception of complex illusions such as the Kanizsa illusion is preserved in patients with hemineglect and extinction. From the experiments presented in this section, it can be concluded that processing of the stimuli presented to the neglected hemifield goes beyond unconscious processing that takes place during CFS or masking. Thus, processing of neglected stimuli is not abolished during early stages, but involves higher level, more elaborated processing.
6. Neural correlates
In the previous sections behavioral studies comparing unconscious processing in healthy subjects and in neglect patients were presented. An important difference between unconscious processing and neglect is that in neglect, patients can become aware of stimuli if their attention is directed to them, whereas in unconscious processing during masked priming subjects are unaware of stimuli despite of spatial attention directed toward the stimuli. How can this difference be explained in terms of brain areas and the mechanisms involved? What are the neural correlates of unconscious processing and neglect?
6.1. Neural correlates of unconscious processing.
Which part of the brain is responsible for conscious perception? A number of neuroimaging studies have contrasted conscious and non-conscious visual processing, but their results appear inconsistent: some support a correlation of conscious perception with early occipital areas, others with late parieto-frontal activity. To resolve these contradictory results, most researchers propose that conscious perception is the outcome of the interactions between brain areas rather than the activation of a single brain region. The neural correlate of consciousness (NCC) is then not restricted to a certain brain area, but can be linked to certain processing stages in different areas.
Dehaene, Changeux, Naccache, Sackur and Sergent (2006) propose that consciousness requires strong bottom-up sensory activation in the presence of top-down attention. Furthermore, in order to reach conscious processing, brain activation must expands into a global parietofrontal reverberant state. The reverberating activation is necessary because it holds information “on-line” for a long duration independently from the initial stimulus duration. In this framework, unconscious processing of masked stimuli corresponds to feedforward, bottom-up activation, which is confined to a brief travelling pulse of firing that is not sufficiently strong to evoke durable fronto-parietal activity. Nevertheless, this activation can reach a semantic level, and result in a short-lived priming effect. Dehaene et. al (2006) stress the importance of higher association cortices in conscious perception, inspired by neuroimaging studies showing that strong activation of higher associative cortices, in particular parietal, prefrontal and anterior cingulate areas, differentiate masked versus unmasked processing of stimuli (Dehaene, Naccache, Cohen, Bihan, Mangin, Poline and Rivière, 2001). Other researchers, however, suggest that a convergence of information towards the prefrontal cortex in itself is not sufficient for conscious sensations to arise (Lamme, 2010). Several studies showed activation of regions such as the frontal eye fields, anterior cingulate, pre-supplementary motor area, inferior frontal gyrus, and anterior insula by masked stimuli or other unconscious events (Lau and Passingham, 2007; van Gaal, Ridderinkhof, Fahrenfort, Scholte and Lamme, 2008).
According to a model proposed by Lamme (2010), recurrent processing and feedback projections are the conditions which are necessary for conscious perceptions to occur, whereas unconscious processing is a result of feedforward processing results. Attention, on the other hand, is determined by the depth of processing. Consequently, two orthogonal divisions are introduced: un-attended perception (shallow processing) vs. attended perception (deep processing), and conscious perception (recurrent processing) vs. unconscious perception (feedforward processing). When combining the depth of the processing with the feedforward vs. recurrent processing, this model distinguishes four stages of processing that can be reached by a stimulus. Stage 1 is characterized by shallow feedforward processing, which activates only lower visual areas ends very early in visual hierarchy. In stage 2, spatial attention deepens processing during the feedforward sweep. This type of processing is characteristic for stimuli that is attended but masked. Stage 3 consists of recurrent but unattended, and thus shallow, processing. In this stage the recurrent processing is restrained to lower brain areas and can create phenomenal experience but lacks reportability. In stage 4 however, attention is brought into play and boosts the processing to be spread to the frontal areas. This deepens the recurrent processing and results in a state of global neural access that can be accompanied by reportability.
Unlike in the model proposed by Deheane, unconscious processing during masked priming is not explained by a lack of strong activation of higher associative cortices. During backward masking unconscious processing can reach higher frontal areas, which is consistent with the results of behavioral and neuroimaging studies which show that unconscious information can be processed in a flexible and adaptive manner (Wokke et al., 2011). In the framework proposed by Lamme (2010), consciousness requires recurrent processing with feedback projections, which is absent in backward
masking. Masked priming hinders recurrent processing of a prime stimulus for the following reason: at the time when feedback projections about a prime would reach early visual areas, activity in these early areas is already overwritten by a mask signal. Since a mask appears later than a prime, feedforward processing is not interrupted and can be additionally boost by spatial attention. When the prime and target share the same properties, the feedforward sweep can pre-activate the sensorimotor pathways related to processing of a target. This results in faster and more efficient processing of a target, i.e. priming.
6.2. Neural correlates of neglect
Neglect is caused by damage to the parietal lobe, usually in the right hemisphere. The brain regions that can define the core anatomy of neglect are the sylvian fissure separating parietal and temporal lobes, the superior and middle temporal cortex with underlying insula, and the ventrolateral prefrontal cortex (Karnath and Rorden, 2012). However, a large variability can be found and a number of different brain regions have become associated with spatial neglect. Since behavioral symptoms of neglect are not always consistent and the degree of impairment may vary between patients, this heterogeneity of brain lesions may be a key factor in determining the diversity of functional deficits observed in this condition. Neglect can also be explained on basis of the two streams hypothesis (Driver and Mattingley, 1998). Two visual pathways can be distinguished: the ventral stream which travels to the temporal lobe and is involved with object identification, and the dorsal stream which projects from primary visual cortex via extra-striate regions into the upper regions of the parietal lobe and is thought to be responsible for the spatial control of action. In neglect, ventral stream is usually intact, while the dorsal stream is damaged.
How can neglect be understood within the frameworks of Dehaene and Lamme? In the model proposed by Dehaene (2006), higher fronto-parietal areas are necessary for conscious perception. Since in neglect these brain regions are damaged, processing stays at a “preconscious” level that is characterized by strong activation, confined to sensori-motor and occipito-temporal areas. This processing involves local resonant firing loops, but top-down attention, necessary for consciousness, is focused on another stimulus or task.
On the basis of Lamme’s theory, parietal damage and attentional deficit hinder deep recurrent processing necessary for global access (stage 4). This causes behavioral deficits in neglect and lack of reportability. On the other hand, considerable residual processing present in neglect can be explained in terms of local or shallow recurrent processing (stage 3).
Evidence for recurrent processing in patients with neglect comes from the processing of visual illusions. Research presented in section 5, suggests that conscious processing of spatial context is necessary for the Kanizsa illusion to occur and that high-level inferential mechanisms are involved. Furthermore, results from various neuroimaging studies show that both feedforward signals and feedback projections play a crucial role in the formation of Kanizsa illusory contours (Seghier and Vuilleumier, 2006). Since a number of experiments demonstrate the processing of illusory contours by patients with neglect (Mattingley et al., 1997; Vuilleumier and Landis, 1998; Vuilleumier et al.,
2001, Conci et al., 2009), taken together, these results strongly suggests that recurrent processing is present in neglect.
6.3. Discussion
In both models described above, neglect and masking are explained by different neural mechanisms. Unconscious processing during masking is presented as a result of feedforward projections (Lamme, 2010) or weak bottom-up activation (Dehaene, 2006). On the other hand, neglect involves recurrent processing with feedback projections (Lamme) and local resonant firing loops (Dehaene).
While there is a general consensus that manipulations such as masking abolish conscious processing, interpretations of neglect involves some controversy. In Dehaene’s framework, attention is necessary for conscious perceptions, and because of its absence neglect is characterized as a “preconscious” state. In the model by Lamme, shallow recurrent processing which characterizes neglect may be sufficient for evoking phenomenal experience (Lamme, 2006, 2010). Lamme proposes a distinction between “phenomenal consciousness” (P-consciousness) and “access consciousness” (A-consciousness). P-consciousness involves phenomenal experience and „raw“ sensation of stimuli, and is characterized by a local recurrent processing (stage 3). A-consciousness requires deep recurrent processing (stage 4) and is linked to attention and reportability. P-consciousness is prior than necessary for A-consciousness.
Thus, neglect may be understood as a problem with A-consciousness but not with P-consciousness, whereas masking interferes with both A- and P-consciousness. Other researchers, who reject the idea of phenomenal consciousness preceding access and reportability, see neglect as a deficit of conscious experience.
7. General Discussion.
In this review I attempted to answer the question of whether processing of neglected stimuli is restrained to early stages or if it involve elaborate processing.
I presented the evidence from various studies showing the covert processing in hemineglect. In patients with neglect, emotional stimuli can in some cases overcome the attentional deficit, in contrast to neutral stimuli. This is the case even when emotional and neutral stimuli share the same low-level characteristics (Vuilleumier and Schwartz, 2001). Faces with emotional expression of fear may be better detected than neutral ones even if the low-level characteristics are the same (Lucas and Vuilleumier, 2008). Furthermore, the same study suggests that gender can be pre-attentively interpreted from faces presented to the contralateral side by neglect patients. Thus, it can be argued that pre-attentive processing of neglected stimuli can result in higher-level representations. Future studies should investigate this issue more in depth. A question that should be addressed is whether gender could be correctly interpreted from a neglected face stimulus without attracting attention. Moreover, findings from semantic priming studies suggest that unconscious semantic processing is possible (Sackur et al. 2008). Even if the number was presented in the neglected field, participants were able to recognize, for example, that ONE and 1 represent the
same number, and that ONE and 4 fall on the same side of 5. This type of reasoning goes beyond low-level information. The evidence also shows that perception of complex illusions is preserved in patients with neglect. From the experiments by Vuilleumier, Valenza and Landis (2001), it can be concluded that processing of the stimuli presented to the neglected hemifield is carried on to the later stages in the visual hierarchy.
All these findings unequivocally indicate that processing of neglected stimuli is not abolished during early stages, but involves higher level, more elaborate processing. The evidence shows that unconscious residual processing can occur within intact brain regions, despite neglect and extinction. Is this type of processing different than the unconscious processing of invisible stimuli by healthy subjects?
Unconscious processing has traditionally been assumed to operate on the lower levels. It was characterized as automatic, pre attentive, operating on simple associations, e.g. motor preparation. In contrast, conscious processing is thought to be dynamic, flexible, adaptive, context specific, elaborated and goal directed. However, new evidence suggests that unconscious processing may have many characteristics that were traditionally attributed to conscious processing. For example, Wokke et al. (2011) showed that unconscious processing can be processed in a flexible, adaptive and context-specific way. Furthermore, emotional stimuli may automatically capture attention outside of awareness and facilitate the unconscious processing which may be carried on to the later stages. The evidence shows that healthy subjects can discriminate between gender if stimuli are sexually arousing (Jiang et al. 2006), whereas for the neutral faces there is evidence for unconscious adaptation modulated by spatial attention (Shin et al. 2009).
What are the differences between semantic processing in neglect and unconscious processing of invisible stimuli by healthy subjects? The measures in the presented studies were mostly behavioral, and it can be argued that behavioral measures may not capture the differences in processing. However, these difference can be understood in terms of brain areas and the mechanisms involved. What are the neural correlates of unconscious processing and neglect? In two models described in section 6, neglect and masking are explained by different neural mechanisms. Unconscious processing during masking is presented as a result of feedforward projections (Lamme, 2010) or weak bottom-up activation (Dehaene, 2006), whereas neglect involves recurrent processing: feedback projections and resonant firing loops.
On the basis of Lamme’s theory, masking is limited to feedforward projections, while considerable residual processing present in neglect can be explained in terms of recurrent processing. Likewise, perceiving complex visual illusions such as the Kanizsa illusion by patients with neglect indicates that processing of the stimuli presented to the neglected hemifield goes beyond feedforward processing that takes place during CFS or masking. Results from neuroimaging studies show that recurrent processing is necessary for the formation of Kanizsa illusory contours (Seghier and Vuilleumier, 2006), as well as the awareness of inducing elements (Harris et al. 2011).
In summary, the studies gathered in this review support the claim that perception of neglected stimuli goes beyond low-level processing and is not restricted to early stages. Similar behavioral findings were presented for unconscious processing during e.g. masking. While there is a general consensus that manipulations such as masking abolish conscious perception, interpretations of neglect involves some controversy.
References:
Adolphs, R., Tranel, D., Damasio, H. and Damasio, A.R. (1994). Impaired recognition of emotion in facial expressions following bilateral damage to the human amygdala. Nature, 372, 669-672.
Amihai I, Deouell L, Bentin S. (2011) Conscious awareness is necessary for processing race and gender. Conscious Cogn. 20(2):269-79.
Anderson, E., Siegel, E. H., White, D. and Barrett, L. F. (2012). Out of sight but not out of mind: Unseen affective faces influence evaluations and social impression. Emotion. doi: 10.1037/a0027514
Berti, A., Rizzolatti, G. (1992) Visual Processing without Awareness: Evidence from Unilateral Neglect. J Cogn Neurosci. 4, 345-51.
Conci M, Böbel E, Matthias E, Keller I, Müller HJ, Finke K. (2009) Preattentive surface and contour grouping in Kanizsa figures: evidence from parietal extinction. Neuropsychologia. 47(3), 726-32. Dehaene,S, Changeux, J.-P, Naccache, L,Sackur, J.,Sergent, C.(2006) Conscious, preconscious and subliminal processing: a testable taxonomy. Trends Cogn. Sci., 10 (5), 204-211.
Dehaene, S., Naccache, L., Cohen, L., Bihan, DL., Mangin, JF., Poline, JB. and Rivière, D. (2001) Cerebral mechanisms of word masking and unconscious repetition priming. Nat. Neurosci. 4, 752–758 Dehaene, S., Naccache, L., Le Clec'H, G., Koechlin, E., Mueller, M., Dehaene Lambertz, G., van de Moortele, P. F., et al. (1998). Imaging unconscious semantic priming. Nature, 395(6702), 597–600. Dimberg, U. (1982). Facial reactions to facial expressions. Psychophysiology, 19, 643–647.
Dimberg, U., Thunberg, M., Elmehed. K, (2000). Unconscious facial reactions to emotional facial expressions. Psychological Science, 11 (1), 86-89.
Driver, J., and Mattingley, J. B., (1998) Parietal Neglect and visual Awareness. Nature Neuroscience, Vol. 1 (1998), pp. 17-22.
Harris, J. J., Schwarzkopf, D. S., Song, C., Bahrami, B., & Rees, G. (2011). Contextual Illusions Reveal the Limit of Unconscious Visual Processing. Psychological Science, 22(3), 399–405.
Haynes, J.D. et al. (2005) Visibility reflects dynamic changes of effective connectivity between V1 and fusiform cortex. Neuron 46, 811–821.
Lamme, V. A. (2010). How neuroscience will change our view on consciousness. Cognitive Neuroscience, 1 (3), 204-220.
Lamme, V. A. F. (2006). Towards a true neural stance on consciousness. Trends in Cognitive Sciences, 10, 494–501.
Lau, HC, Passingham, RE. (2007) Unconscious activation of the cognitive control system in the human prefrontal cortex. J Neurosci 27, 5805–5811.
Lucas, N., Vuilleumier P. (2008) Effects of emotional and non-emotional cues on visual search in neglect patients: Evidence for distinct sources of attentional guidance. Neuropsychologia. 46(5), 1401-14.
Marshall, J. C., Halligan, P. W. (1988) Blindsight and insight in visuo-spatial neglect. Nature, 336(6201):766-7.
Mattingley J. B., Davis, G., Driver, J. (1997) Preattentive filling-in of visual surfaces in parietal extinction. Science, 31; 275 (5300), 671-4.
Morris J. S., Ohman, A., Dolan, R. J. (1998) Conscious and Unconscious Emotional Learning in the Human Amygdala. Nature 393, 467-470.
Murphy, S. T.; Zajonc, R. B. (1993) Affect, cognition, and awareness: Affective priming with optimal and suboptimal stimulus exposures. Journal of Personality and Social Psychology, 64(5), 723-739. Naccache L, Dehaene S. (2001) Unconscious semantic priming extends to novel unseen stimuli. Cognition, 80, 215-229.
Neumann, O., and Klotz, W. (1994). Motor responses to nonreportable, masked stimuli: Where is the limit of direct parameter specification? In C. Umiltá & M. Moscovitch (Eds.), Conscious and unconscious information processing (123-150). Cambridge, MA: MIT Press.
Odgen, J. (2012) Trouble in Mind. OUP, New York.
Sackur J, Naccache L, Pradat-Diehl P, Azouvi P, Mazevet D, Katz R, Cohen L, and Dehaene S: Semantic processing of neglected numbers. Cortex 2008, 44, 673-682.
Seghier, M., Vuilleumier, P. (2006) Functional neuroimaging findings on the human perception of illusory contours. Neuroscience and Biobehavioural Reviews 30, 595-612.
Shin K., Stolte M., Chong S. C. (2009). The effect of spatial attention on invisible stimuli. Atten. Percept. Psychophys. 71, 1507–1513
van Gaal, S., Ridderinkhof, K. R., Fahrenfort, J. J., Scholte, H. S., & Lamme, V. A. F. (2008). Frontal cortex mediates unconsciously triggered inhibitory control. Journal of Neuroscience, 28, 8053–8062. Vuilleumier, P., Landis, T. (1998). Illusory contours and spatial neglect. NeuroReport 9, 2481–2484.
Vuilleumier P, Valenza N, Landis T. (2001) Explicit and implicit perception of illusory contours in unilateral spatial neglect: behavioural and anatomical correlates of preattentive grouping mechanisms. Neuropsychologia, 39(6), 597-610.
Vuilleumier P. Armony, J., Clarke, K., Husain, M., Driver, J., & Dolan, R. (2012) Neural response to emotional faces with and without awareness: event-related fMRI in a parietal patient with visual extinction and spatial neglect. Neuropsychologia, 40(12), 2156.
Webster, M. A., Kaping, D., Mizokami, Y., & Duhamel, P. (2004). Adaptation to natural face categories. Nature, 428, 557–561.
Wokke, M., van Gaal, S., Scholte, H. S., K. Richard Ridderinkhof, K. R., Lamme, V. A. F. (2011) The Flexible Nature of Unconscious Cognition. Plos ONE, 6(9)
