Hulleman & Olivers (H&O) have given a deeply convincing case for why we should consider fixations, and not items, as the fundamental unit of search. Central to their framework is the Functional View Field (FVF), which is an area of the visual field centred on fixation, where an item can be expected to be detectable. Towards the end of their article, H&O pose the question of what factors determine the size of the FVF. One important question they did not ask is whether the FVF can change shape, and if so, under what circumstances. Moreover, H&O leave open the possibility of grouping mechanisms within the FVF. Here I consider how findings relating to object-based attention provide possible answers to these questions.
To begin, it is accepted that visual attention has both spatial- and object-based components (Soto & Blanco Reference Soto and Blanco2004). Spatially based visual attention involves attention being directed to a general area of the visual scene (Downing Reference Downing1988). Object-based visual attention (OVA), on the other hand, is directed toward objects or groups of elements adjoined by Gestalt principles (Neisser Reference Neisser1967), and it has been supported in part by findings that, all else being equal, RTs are faster when a cue and target appear within the same object compared to when they are not (Egly et al. Reference Egly, Driver and Rafal1994). While there is an obvious spatial component to visual search, it is interesting to consider how, in the context of H&O's framework, the FVF may process information given what is currently known about object-based attention.
Evidence for intra-FVF grouping can be tentatively drawn from studies using the aforementioned cued-rectangle paradigm. For example, Norman et al. (Reference Norman, Heywood and Kentridge2013), report an object-based effect with rectangles that were not conscious to the viewer. Viewers fixated on a point, then made a timed response to a target stimulus preceded by a cue. As with previous findings using this paradigm (e.g., Egly et al. Reference Egly, Driver and Rafal1994), viewers were faster to detect the target when it appeared within the same object as the cue. Applied to H&O's framework, this would suggest that once a fixation is made at a given location, such as on a cue, the FVF's contents are not processed uniformly (as H&O readily admit); rather, they are subject to grouping mechanisms that prioritise some visual elements within the FVF over others.
There is, however, disagreement over the mechanisms of object-based attention. Some argue that several OVA effects exemplify attentional prioritization of locations that simply assigns a higher probability of the target appearing within the boundaries of an attended object (Shomstein & Yantis Reference Shomstein and Yantis2002; Greenberg et al. Reference Greenberg, Rosen, Cutrone and Behrmann2015), not unlike priority of scene locations based on local contrast (e.g., Wolfe Reference Wolfe2010). Thus, any possible grouping in the visual scene may facilitate search, but only because it guides fixations, not because of any alteration of the FVF. As a result my interpretation of Norman and colleagues' (2013) findings remains tentative, as they did not account for eye-movement during their task.
How might OVA influence the shape of the FVF? One way is by sensory enhancement (Desimone & Duncan Reference Desimone and Duncan1995). Imagine that the “default” FVF during visual search of a given difficulty is a circular patch with fuzzy boundaries. When there are no sufficiently strong grouping factors between visual elements, such as in most search arrays, only the contents of the circular FVF are processed. However, if a peripheral visual element is grouped with an element within the “default” FVF, the signal from the peripheral visual element is “boosted” and is subsequently processed more thoroughly than it would have been otherwise, effectively expanding the FVF. Wannig et al. (Reference Wannig, Stanisor and Roelfsema2011) provide evidence to support such an account. They recorded two receptive fields (area V1) of monkeys during a visual task. The monkeys were trained to foveate on a target line stimulus, and during the task the target line was flanked by task-irrelevant line stimuli. After fixation on the task-relevant stimulus, the authors found an increase in activity from the receptive field corresponding to a task-irrelevant line stimulus, but only if the task-irrelevant stimulus was collinear with the task-relevant stimulus. This demonstrates that: (1) attentional spreading takes place in early representations of visual elements grouped by Gestalt factors; (2) this attentional spreading occurs automatically (i.e., attention spreads to task-irrelevant stimuli). Further experimentation will be needed to see whether this kind of attentional spread has any bearing on visual search.
H&O remarked that their framework has no problem explaining search across a messy scene bereft of clearly-defined items. I agree, but I also stress that there is still much room within this framework for a scene to be divided into object-like groupings during search. It would be out of neglect for what is known about object-based attention to assume that the FVF always processes information irrespective of object groupings. Furthermore, although retinal physiology is no doubt crucial in determining the boundaries of the FVF, we have good reason to believe that these boundaries are also subject to early “signal boosting” as a result of Gestalt grouping.
You have
Access
Hulleman & Olivers (H&O) have given a deeply convincing case for why we should consider fixations, and not items, as the fundamental unit of search. Central to their framework is the Functional View Field (FVF), which is an area of the visual field centred on fixation, where an item can be expected to be detectable. Towards the end of their article, H&O pose the question of what factors determine the size of the FVF. One important question they did not ask is whether the FVF can change shape, and if so, under what circumstances. Moreover, H&O leave open the possibility of grouping mechanisms within the FVF. Here I consider how findings relating to object-based attention provide possible answers to these questions.
To begin, it is accepted that visual attention has both spatial- and object-based components (Soto & Blanco Reference Soto and Blanco2004). Spatially based visual attention involves attention being directed to a general area of the visual scene (Downing Reference Downing1988). Object-based visual attention (OVA), on the other hand, is directed toward objects or groups of elements adjoined by Gestalt principles (Neisser Reference Neisser1967), and it has been supported in part by findings that, all else being equal, RTs are faster when a cue and target appear within the same object compared to when they are not (Egly et al. Reference Egly, Driver and Rafal1994). While there is an obvious spatial component to visual search, it is interesting to consider how, in the context of H&O's framework, the FVF may process information given what is currently known about object-based attention.
Evidence for intra-FVF grouping can be tentatively drawn from studies using the aforementioned cued-rectangle paradigm. For example, Norman et al. (Reference Norman, Heywood and Kentridge2013), report an object-based effect with rectangles that were not conscious to the viewer. Viewers fixated on a point, then made a timed response to a target stimulus preceded by a cue. As with previous findings using this paradigm (e.g., Egly et al. Reference Egly, Driver and Rafal1994), viewers were faster to detect the target when it appeared within the same object as the cue. Applied to H&O's framework, this would suggest that once a fixation is made at a given location, such as on a cue, the FVF's contents are not processed uniformly (as H&O readily admit); rather, they are subject to grouping mechanisms that prioritise some visual elements within the FVF over others.
There is, however, disagreement over the mechanisms of object-based attention. Some argue that several OVA effects exemplify attentional prioritization of locations that simply assigns a higher probability of the target appearing within the boundaries of an attended object (Shomstein & Yantis Reference Shomstein and Yantis2002; Greenberg et al. Reference Greenberg, Rosen, Cutrone and Behrmann2015), not unlike priority of scene locations based on local contrast (e.g., Wolfe Reference Wolfe2010). Thus, any possible grouping in the visual scene may facilitate search, but only because it guides fixations, not because of any alteration of the FVF. As a result my interpretation of Norman and colleagues' (2013) findings remains tentative, as they did not account for eye-movement during their task.
How might OVA influence the shape of the FVF? One way is by sensory enhancement (Desimone & Duncan Reference Desimone and Duncan1995). Imagine that the “default” FVF during visual search of a given difficulty is a circular patch with fuzzy boundaries. When there are no sufficiently strong grouping factors between visual elements, such as in most search arrays, only the contents of the circular FVF are processed. However, if a peripheral visual element is grouped with an element within the “default” FVF, the signal from the peripheral visual element is “boosted” and is subsequently processed more thoroughly than it would have been otherwise, effectively expanding the FVF. Wannig et al. (Reference Wannig, Stanisor and Roelfsema2011) provide evidence to support such an account. They recorded two receptive fields (area V1) of monkeys during a visual task. The monkeys were trained to foveate on a target line stimulus, and during the task the target line was flanked by task-irrelevant line stimuli. After fixation on the task-relevant stimulus, the authors found an increase in activity from the receptive field corresponding to a task-irrelevant line stimulus, but only if the task-irrelevant stimulus was collinear with the task-relevant stimulus. This demonstrates that: (1) attentional spreading takes place in early representations of visual elements grouped by Gestalt factors; (2) this attentional spreading occurs automatically (i.e., attention spreads to task-irrelevant stimuli). Further experimentation will be needed to see whether this kind of attentional spread has any bearing on visual search.
H&O remarked that their framework has no problem explaining search across a messy scene bereft of clearly-defined items. I agree, but I also stress that there is still much room within this framework for a scene to be divided into object-like groupings during search. It would be out of neglect for what is known about object-based attention to assume that the FVF always processes information irrespective of object groupings. Furthermore, although retinal physiology is no doubt crucial in determining the boundaries of the FVF, we have good reason to believe that these boundaries are also subject to early “signal boosting” as a result of Gestalt grouping.