Changing fixation patterns in line with an instruction

In document Measuring the bias to look to the eyes in individuals with and without autism spectrum disorder (Page 42-47)

Chapter 1: Introduction

1.5 Changing fixation patterns in line with an instruction

As has been described, social stimuli, especially the eyes, appear to capture attention and elicit greater recognition for NT individuals. In addition, although the findings are mixed in ASD, many studies suggest ASD individuals do look to the eye region and that their early fixations to the eyes are automatic. Understanding the orientation of automatic looking in autistic individuals is important as it helps us to understand any biases that exist in ASD and whether or not this is to social stimuli. Increased information on this automatic bias would enable further understanding of the symptom presentation in ASD and the drives behind these symptoms, i.e. having a bias to look to social stimuli but inhibiting this drive in daily life, or not having a bias to look at social stimuli in the first place. Therefore understanding whether or not there is an automatic bias to look to the eyes in autistic individuals would enable further understanding of those with autism and what drives any difficulties with social interaction they may experience in daily life. To enable further understanding of how automatic these fixation patterns are, a small number of studies have attempted to manipulate participants’ gaze direction through a range of different paradigms to see if it is possible to change these face processing patterns.

1.5.1 Changing fixation patterns in NT individuals

A variety of paradigms have been used to investigate how successfully NT participants’ gaze can be altered when viewing face stimuli in a range of different situations. One way of investigating this is by using magic tricks that normally rely on being able to predict where a person will automatically look. In an experimental manipulation, natural looking patterns, which typically lead to the viewer being tricked, can be compared to looking patterns when the individual is instructed on where to look. In one study it was found that an instruction to look to an alternative location, rather than the eyes, had some success, but the adult participants were unable to fully follow this instruction (Kuhn, Teszka, Tenaw, & Kingstone, 2016). In this paradigm participants watched a card trick and were told their task was to work out how it was done. Half of the participants were explicitly told to keep looking at the cards, as magicians often use misdirection, and the other half were not given any further instruction. Participants who were given this explicit instruction did fixate more on the cards and looked to the face less than those that received no instruction. However, despite this explicit instruction less than 5% of the participants instructed to only look at the cards managed not to look at the face at all during the interaction. This, therefore, suggests that the automatic and initial tendency to fixate on the face, is difficult to fully overcome and can only be partially moderated by explicit instruction.

The findings in this magic trick study are supported by the findings in a look/ don’t look paradigm, which also found that adults had difficulty following instructions to not look at the eyes (Laidlaw et al., 2012). The participants completed a “don’t look” (DL) task, where half were to avoid looking to the eyes (DL: eyes) and the other half the mouth (DL: mouth) for the five second presentation time. Although participants in both DL conditions reduced their fixation count and dwell time to the eyes and mouth respectively, those in the DL: eyes condition made significantly more errors (looked to the eyes significantly more) than those in the DL: mouth condition. In addition, when comparing the fixation count and dwell time to the amount of looking that would be expected if viewing was random across the screen, those in the DL: eyes condition looked more to the eyes than would be expected by chance, however, those in the DL: mouth condition looked to the mouth less than would be expected by chance. The authors also divided up the viewing time into the first second (early viewing period) and the remaining viewing time (late viewing period) to compare if inhibiting looking

was easier to control with more time available to participants. They found that the amount of errors regarding DT for the DL: eyes condition, was significantly greater in the early viewing period, compared to the late viewing period, however both of these viewing periods resulted in more time viewing the eye region than would be expected by chance. This same effect was not seen in the DL: mouth condition, where there was no difference in the DT errors for the early and viewing periods.

Although the previous studies present a clear picture of the difficulties in not looking to the eyes, either when explicitly instructed not to or when instructed to look elsewhere, a study conducted by Itier et al. (2007) presents a slightly more complex picture. The authors attempted to manipulate adults’ gaze patterns to a static face stimulus by providing the participants with two different tasks, one that required them to fixate the eyes and one that did not. Firstly, participants completed a gaze direction task, where they had to state whether the person in the image was looking at them or away. Secondly, they completed the head direction task where they had to distinguish between the images where the head was straight on and those where the head was in a ¾ view. In this second task participants were explicitly told they did not need to look at the eyes. It was found that there was significantly higher dwell time for the eye region in the gaze condition, compared to the head condition, however, there was more looking to the eye region than the lower face in both of these conditions. In addition, the authors used the measure of first fixations to understand whether participants would always look to the eye region first, and therefore that eye looking was reflexive and automatic, or whether this eye looking depended on the task being completed. They found that there were more initial saccades made to the eye area than the lower face area in the gaze task. However there were no differences between the number of eye and mouth first saccades in the head condition. The authors

conclude that although there is a bias to look to the eyes in both tasks, this bias to initially saccade to the eye region appears to be task dependent.

These three different tasks all try to understand how successfully a NT adult’s gaze can be changed in line with either a direct or an indirect instruction. They all show that, to some extent, there is a bias to look to the face or the eyes and that this looking still occurs even when participants are told not to or told it will not help them to complete the task. Laidlaw et al. (2012) propose that this difficulty to fully look away from the eyes even when told to do so represents an automatic component of eye

looking, which is not under volitional control. These previous studies could also be considered within a top down and bottom up theoretical framework of attention. Top down attention is the process of voluntarily allocated attention to look to certain objects or features, whereas bottom up attention is not voluntarily directed and is the result of salient stimuli attracting attention (Carrasco, 2011; Corbetta & Shulman, 2002; Desimone & Duncan, 1995; Ungerleider & G, 2000). Therefore, these previous studies could suggest that although the majority of face looking is under volitional control, or with top down attention being dominant, there does appear to be a bottom up attention component that is more automatic. This is characterised by the difficulty in fully inhibiting looking to the eyes in the don’t look paradigm (Laidlaw et al., 2012), the low number of participants who managed to not look to the eyes during the magic trick (Thomas & Didierjean, 2016) and the bias to look to the eyes in Itier et al.’s task.

These three previous paradigms considered together suggest that it is possible to reduce eye looking to some extent in line with an instruction, indicating there is some control over looking patterns, but the eyes do seem to draw attention and this is partly an automatic orientation that is not possible to fully overcome.

1.5.2 Changing fixation patterns in ASD individuals

As has been shown, some of the face processing fixation patterns in the NT population do appear to be difficult to change even when explicitly instructed to do so. This difficulty in consciously changing the way people look at faces may suggest that, to some extent, face processing patterns may be automatic, with people being drawn to the eye areas and struggling to change this even when explicitly told to do so. In ASD, however, there are many varied findings regarding face processing patterns in the previous research. This variation may suggest that the patterns of dwell time vary depending on the task and the paradigm in question. Subsequently, it could be argued that autistic individuals may not have such dominant automatic processing patterns and that their patterns of eye gaze may be easier to change. Alternatively, it could be that there are subgroups within ASD and that the variety seen is not as a result of task demand and if this is the case it is less clear how easy the face processing patterns of these groups would be to change.

There have been a limited number of studies investigating changing looking patterns towards faces in those with ASD and these studies use different paradigms

and have drawn varying conclusions. One study used a gaze following task (as outlined in section 1.4.4, see page 26), but with a cued condition whereby children were instructed to identify the object being looked at (Riby, Hancock, Jones, & Hanley, 2013). The authors found that the ASD participants did not automatically look to the eyes and follow the gaze in the spontaneous condition, but the participants also did not change their gaze patterns to do this in the cued condition either. This suggests the ASD participants were not able to alter their looking patterns in line with an instruction in this paradigm. In contrast to this, Kikuchi et al. (2011), using a gap overlap task where a face is the central distractor, found that although ASD children did not look automatically to the eyes they were able to alter their looking patterns to look at the eyes when instructed. Different findings were found in a gaze cueing task where the eyes were looking at objects on either side of the screen (Rombough & Iarocci, 2013). The children were told whether they should look to the eyes to help them to complete the task or whether they should avoid looking to the eyes. In this task both the ASD and TD groups looked to the eyes regardless of the instruction and as a result showed poorer performance on the trials where the gaze direction was invalid and therefore the eyes were only being used as a distractor.

In summary, only a limited number of studies have investigated changing face gaze patterns and a wide range of methodologies have been used. The NT studies (Kuhn et al., 2016; Laidlaw et al., 2012) suggest that it is difficult for people to look away from the eyes even when explicitly instructed to do so. This may suggest that people automatically orient to the eye region and this may be out of their top down attentional control and is instead a bottom up process. However, the picture in the ASD studies is not as clear, with one study suggesting the participants were able to alter their looking patterns in line with a prompt (Kikuchi et al., 2011), but two suggesting difficulty altering face gaze patterns, either a difficulty in orienting to the eyes under instruction (Riby et al., 2013) or difficulty orienting away from the eyes under instruction (Rombough & Iarocci, 2013). There is not a clear picture across these studies of whether looking to the eyes is a bottom up or top down process for those participants with ASD, as the findings are contradictory. This again highlights the varied results seen in ASD research and the importance of trying to understand face gaze patterns further. This is especially important in ASD due to the social and

understand whether those with ASD can alter their looking patterns when looking at a face will help to establish whether there is an automatic bias to look to social stimuli (i.e. the eyes) which could result in further understanding of the behaviours of those with ASD and enable further understanding and support for the reasons behind these behaviours.

In document Measuring the bias to look to the eyes in individuals with and without autism spectrum disorder (Page 42-47)