History of eye-tracking in psychological research In the 1950s, Alfred Yarbus showed the task given to a subject has a very large influence on the subjects eye movements. Yarbus also wrote about the relation about fixations and interest. This study by Yarbus (1967) is often referred to as evidence on how the task given to a person influences his or her eye movements. In the 1970s, eye tracking research expanded rapidly, in particular reading research. A good overview of the research in this period is given by Rayner. In 1980 Just and Carpenter formulated the influential so-called strong eye-mind hypothesis which states that "there is no appreciable lag between what is fixated and what is processed". It is a hypothesis that is today much too often taken for granted by many beginning eye tracker researchers.
During the 1980s, the eye-mind hypothesis was often questioned, in the light of covert attention, i.e. you can attend to something different than what you are looking at. If covert attention were to be common during eye tracking recordings, the resulting scan path and fixation patterns would often show not where our attention has been, but only where the eye has been looking. Current consensus is that visual attention is always slightly - between 100 and 250 ms - ahead of the eye. But as soon as visual attention moves to a new position, the eyes will want to follow. Today eye tracking data permit inferences on cognitive processes (e.g. recognition, liking, dislike, puzzlement, etc) directly from a fixation on a particular object in a scene. Eye tracking can be coupled with other methodologies, for instance retropective verbal protocols.
Gaze: Perception & Attention Tasks and paradigms in eye movements research Concetta Alberti
Fixations: Information encoding The perceptual span Next slides Guidance of saccades The latency of saccades The landing position of saccades Attention Voluntary and automatic saccades Covert attention
-Fixations- 3 processes at least take place during a fixation: 1) visual information is encoded 2) information from the periphery of the visual field is acquired in order to determine subsequent informative areas 3) next saccade is planned and prepared. 2 precess take place during a fixation: (Viviani, 1990) 1) discrimination 2) selection (Hooge & Erkelens, 1996,1998, 1999)
1) Information Encoding During a fixation the information from the fixated area are usually extracted Properties of the fixated area influence fixation durations
2) The perceptual span and the PREVIEW BENEFIT : Information is also partially processed in the periphery of the visual field. The parafoveal information might be processed to different levels: Totally unprocessed form (iconic) Partially processed form (activation of the semantic network, called semantic priming). Partially processed in order to guide the eyes to the subsequent location (landing position).
The perceptual span Also called: stationary field, functional field of view (FFoV), conspicuity area, visual lobe and visual span. It is an area surrounding the point of regard from which information is extracted in the course of a fixation. It is a visual and attentional limit. The size of the perceptual span has been studied in: Psychophysical studies: measured the distance from fixation at which a target can be detected or localized (Sekuler and Ball, 1986). Gaze contingent paradigms: as the eyes move all material around the fixation is seen normally, the rest is modified. Researchers manipulate the size of the window and determine the size at which performance reaches normal levels (McConkie & Rayner, 1975, 1976). Double-task: the span size is reduced by increases in workload (Pomplun et al., 2001). Visual search: the span depends on the similarity of target and distractor, it is narrower when they are more similar (Rayner & Fisher, 1987). Reading : the perceptual span is asymmetrical (3-4 L, 14-15 R).
The perceptual span (Sanders,1963) Stationary field Eye field Head field Response time increases if comparative target is moved from the stationary field to the eye field, and moving it again towards the head field increases RT further more. Integrating visual information across multiple gazes
-Guidance of Saccades- Two dimensions are of interest in the programming of saccades: when and where. When: Latency of the saccade Where: Landing position of the saccade
Guidance of Saccades The process is influenced by: Similarity effects Proximity effects Salience map Similarity effects: saccades are more probably directed to items which are similar to the target Proximity effects: saccades are more probably directed to items which are close to the visual axis (Motter e Belky, 1998). The strength of the proximity effect is influenced by item density or crowding (elements in space unit) and item salience. Salience map: saccades are more probably directed to the most conspicuous items in the map; the map represents the scene and each object in the map compete for selection. When the target is salient proximity effects are overridden
1) THE LATENCY OF SACCADES: THE GAP EFFECT (when) Task: The observer is asked to follow a target which makes unpredictable step movements (disappears at the previously fixated location and appears at the new location) The temporal offset between the events was varied (gap or overlap). Does the interval between disappearance and reappearance influence the programming of the saccade? Saccade latencies become shorter in the gap situation with increasing temporal offsets Saccade latencies become longer with increasing overlap.
2) THE LATENCY OF SACCADES : THE REMOTE DISTRACTOR EFFECT OF LATENCY (when) A new stimulus appears around the time of target onset When the new stimulus appears simultaneously on the controlateral side with respect to the target the latency increases. When the new stimulus appears simultaneously on the ipsilateral side with respect to the target the latency is not influenced but the amplitude of the saccade is modified (Walker et al. 1997). When the stimulus onset is previous to target appearance the latency decreases (Walker et al. 1997). Therefore, the latency (not landing position) is affected by distractors further away, when the distractor is near the target the landing position (not the latency) is affected.
3) The landing position: DOUBLE target PARADIGM and the Global Effect Two targets appear simultaneously. When the targets are in reasonably close proximity the orienting saccade goes to an intermediate location between the targets. Global effect or centre-of-gravity effect (Findlay, 1982; Deubel et al., 1984; Ottes et al., 1984) The relative properties of the targets (size and brightness) influence saccade landing position Landing position (not the latency) of the saccade is influenced by the presence of near distractors
4) THE LANDING POSITION: Accuracy in landing position There are 2 main concerns with landing position: The eye can land on the target or, if not, on a distractor with some characteristics (selection accuracy) The eye can land accurately on an item or between 2 items (targeting accuracy)
-Attention- 3 physical shifts of attention: disengagement, shifting, engagement (Eysenck & Keane, 2005) Prior to an overt eye movement, covert attention shifts to the target location (Hoffman & Subramaniam, 1995; Kowler et al., 1995; Deubel & Schneider, 1996; Peterson, Kramer, & Irwin, 2004).
Voluntary Direction and Automatic Capture of attention. Attention can be directed either voluntarily, also referred to as endogenous control, or automatically, which is also called exogenous or reflexive capture of attention. Endogenous control involves volition to direct attention Exogenous control occurs when an external object or event grabs attention and attracts it involuntarily. The neural mechanisms in the brain have been shown to produce differt patterns of activity for endogenous and exogenous attention (Gazzaniga, et al., 2002).
Voluntary and Reflexive or targetelicited SACCADES Volitional saccades are cognitively demanding and must be elicited by behavioral paradigms: The antisaccade task, originally introduced by Hallet, is an example of voluntary saccade task. Another paradigm used to study voluntary saccades proposes memory-guided after delay saccades Reflexive saccades are elicited by a suddenly appearing visual target that captures attention and triggers a saccade to it.
Covert Attention or Salience map Selection of next saccadic target (landing position) 1) Is a serial scanning operated by covert attention (Peterson et al., 2001)? 2) Is a parallel process in which the most salient item is selected across the visual field?
Failure in eye tracking Lapses in visual encoding: Even if a fixation has occurred on an item of interest, there is no guarantee that it has been processed to the point of recognition or access to working memory. It can be due to visual or nonvisual cognitive load, and be indipendent of gaze duration. Examples: radiological report, driving during mobile conversation, change blindness. On the other hand, a location can be attended although it has not been fixated (covert attention). Human factor specialists claim that probably covert attention doesn t play a significant role in real-world task.
Visual search Visual search The task requires to extract information in subsequent fixations on the screen But the amount of information that can be extracted from a single fixation influences the number of fixations and the visual search time. The amount of information that is extracted in a single fixation is influenced by the width of the perceptual span The perceptual span size depends on different issues: Similarity of the items Proximity of the items General visual load
Visual Search: a model Visual search is guided by attivation within a salience map (Koch & Ullman, 1985) A salience map is the result of low level representations of a scene basic features Contrast between feature values produces bottom-up signals that are fed forward to the salience map. There can be top-down modulation of activation within the feature maps and within the response. It amplifies signals that encode known target properties and attenuate activation within the maps that encode properties that do not belong to the target. During visual search observers preferentially fixate stimuli that share features with the target (Findlay, 1997).
Free viewing of scenes The observer's eye repeatedly returns to the same elements of the picture: additional time spent on perception is not used to examine the secondary elements, but to reexamine the most important elements. The observer's attention is frequently drawn to elements which do not necessarily give important information but which, in his opinion, may do so. Often an observer will focus his attention on elements that are unusual in the particular circumstances, unfamiliar or incomprehensible.
Real-life environments and tasks. Traffic scenes: observer fixate task-relevant object and conspicuous, salient items. Making sandwiches requires a series of actions: ORAs (object related actions) are delineated by a shift of the eyes from the object currently being manipulated to the object next to be manipulated. A fixation on each object begins a fraction of a second before the manipulation of the object is initiated. The proportion of fixations on task-irrilevant objects is <5% (goal-directed eye movements).