Attentional limits and freedom in visually guided action

https://doi.org/10.1016/S0079-6123(09)17612-4Get rights and content

Abstract

Human vision supports both the conscious perception of objects (e.g., identifying a tea cup) and the control of visually guided action on objects (e.g., reaching for a tea cup). The distinction between these two functions is supported by neuroanatomy, neuropsychology, animal lesion studies, psychophysics and kinematics in healthy humans, as well as in overarching theories of the visual brain. Yet research on visual attention, which concerns limitations in processing information concurrently from multiple objects, has so far not fully exploited this functional distinction. Attention research has focused primarily on the conscious aspects of vision, to the relative neglect of the unconscious control of limb actions and whether we can perceive some objects concurrently while acting on others. Here we review research from our lab leading to the conclusions that (1) the finger is guided visually by an automatic pilot that uses different information from that of conscious vision, (2) conscious object identification interferes with concurrent planning of pointing to a second object, though not with the online control needed to complete the pointing action, (3) concurrent perception and action sometimes lead to benefits in motor performance, and (4) the automatic pilot is itself capacity limited in processing information concurrently from multiple locations. These findings help clarify the conditions under which interference-free multitasking is possible and point to new challenges for research on the attentional limits of unconscious visual processing.

Introduction

It is time for research on attention to catch up with recent developments in our understanding of how the human brain uses visual information to perform distinctly different functions. Although there is only one stream of light that impinges on our eyes at any moment in time (hereafter referred to as the image), the information contained in that image can be used to serve one of two functions. On the one hand, the image can be used to construct a visual experience of the environment around us, allowing us to recognize a colleague or to discern whether we are viewing scissors or a spoon. Yet the same image can also guide our limb actions so that we interact appropriately with that environment, allowing us to shake our colleague's hand when we greet them and to pick up scissors using a different grasp than we use when picking up a spoon. Research on human vision has demonstrated in numerous ways over the past 25 years that these two functions are distinct (Goodale and Milner, 2004; Goodale et al., 2004; Milner and Goodale, 1995; Norman, 2002; Ungerleider and Mishkin, 1982). Specifically, our visual experience in the absence of overt action (the conscious experience of objects at a distance, subserved by the ventral visual stream) is governed by brain regions, neural tracts, and mechanisms that are distinct from those that guide our interactions with objects (unconscious control of visually guided action, subserved by the dorsal stream). It is not our purpose to review those arguments here, as they have been made extensively in many reviews. Our purpose is, instead, to consider some consequences of this understanding of the visual brain for research on visual attention.

Research on visual attention over the past 150 years can be characterized, at a first approximation, as the study of channel limitations of the human brain as it concerns conscious experience (see Pashler, 1994, Pashler, 1998; Shapiro et al., 1997 for reviews). We seem to be aware, and to become aware, of little more than one discrete event at a time. If we are expecting imminent visual information from one location, we will be delayed in processing the expected visual information at another location (e.g., location orienting, as in Posner, 1980). If we are engaged in identifying one person, the identification of a second person must wait (e.g., two object cost, as in Duncan, 1984). Even when we are focused solely on viewing a single person's face, if we are engaged in the process of identifying that person, we will be impaired in making judgments of the emotional expression currently displayed by that person (e.g., we are blind to changes that are deemed unlikely, as in Austen and Enns, 2003). If, during the performance of any of these visual tasks, we temporarily allow our thoughts to drift among the ideas constantly being offered to our awareness by our long-term memories, then our ability to identify and localize visual information becomes impaired (e.g., mind wandering, as in Giambra, 1995).

Relative to this intense interest in attentional limits on conscious aspects of seeing, much less effort has been devoted to developing methods for exploring limits of the visually guided action system, and for exploring the conditions under which conscious vision and visually guided action might interfere with one another, might operate independently of one another, or perhaps might even mutually enhance one another. Among the obstacles encountered in the design of experiments to probe unconscious visually guided action is that it is often difficult to provide input to the action system that does not itself have to pass through the bottleneck of consciousness before it can be used to inform the visually guided action system. Take, as a case in point, dual-task studies already in the literature that have combined tasks of action and perception in an effort to measure possible interference when these systems must use the same visual image. Some of these studies required participants to point to one colored shape while simultaneously trying to identify a symbol in a separate location (Deubel et al., 1998). Other studies required participants to grasp a target object while simultaneously monitoring for changes in the luminance of a second object (Castiello, 1996). Since pointing and grasping are thought to be under dorsal stream control and object identification under ventral stream control, these could be construed as proper tests of cross-stream interference. And since significant task interference was observed in both studies, one might conclude that efficient multitasking is not possible between the visual streams. However, we do not consider these results to be strong tests, mainly because in order to carry out the limb action required in both types of studies, the color or the shape of an item has to be processed (ventral stream function) before the appropriate action (dorsal stream function) can be initiated.

A second difficulty encountered when designing experiments to probe unconscious visually guided action separately from conscious vision concerns the important distinction between action planning (or preparation) versus the online control of that action once it is already underway (Henry and Rogers, 1960). Action planning is generally considered to involve processes that occur prior to action initiation, and therefore may be influenced by the ventral stream as well as other consciously accessible brain regions such as the frontal lobe functions implicated in executive task control. It is typically indexed by measuring the period from target onset to movement onset (response initiation time), and as such, is potentially influenced by the mental processes of target identification, response selection, and movement planning (or preprogramming). In contrast, action execution consists of processes involved in bringing the action to completion and is usually considered more uniquely a function of the dorsal stream system. It is indexed by the time that elapses between action onset and action completion (movement time, MT) and is thus uniquely influenced by processes that occur only once an initiated action is already underway.

An example of research that ignores this distinction comes from recent report by Kunde et al. (2007). These authors paired an auditory tone discrimination task (ventral stream) with either a visual size discrimination task (ventral stream) or a visually guided grasping task (ostensibly dorsal stream). Their results indicated that response time in both visual tasks was interfered with when these tasks were paired with the auditory task. Although, the dependent measures taken in the grasping task included both response time (preparation) and MT (online control) only the response time data were reported in detail, leaving open the possibility that interference only occurred during the planning phases of the action. Consistent with this possibility was a terse one-line report by the authors that they found no reliable influences of dual-task performance on MT. But this null result was not explored in any greater detail, even though it suggested the possibility that online control of grasping was immune from the task interference measured for action initiation. Our reading of the existing literature on dual-task performance involving ventral and dorsal stream functions therefore suggests that much remains to be done. In particular, there is a need for research involving dorsal stream tasks that (1) does not rely on consciously processed input for their initiation, and (2) allow for the separate measurement of action planning (or preparation) from the online control of already initiated actions.

Section snippets

A model task for studying online control: the finger's automatic pilot

Pressing an elevator button while we are walking toward the door, reaching to grasp someone's hand as we both move toward one another, and striking a tennis ball that has just tipped the top of the net, are all highly sophisticated visual-motor skills that we as healthy humans take for granted. Many complex computations are involved, though they remain hidden from our ability to access them through conscious introspection. This human ability to make rapid, online adjustments in pointing and

Attention sharing between visual identification and the automatic pilot

The tasks we chose to combine in a dual-task setting involved participants monitoring a rapid-fire sequence of digits for the presence of a central letter target (ventral task) while simultaneously pointing to a second letter target that appeared with variable temporal lag in the visual periphery (dorsal task) (Liu et al., 2008). Additionally, in a single-task condition, participants ignored the central target while pointing to the peripheral target. This is a version of the well-established

Capacity limits of the automatic pilot?

In the study of attentional limitations on conscious perception, it is conventional to refer to a task as automatic when it can be accomplished without interference from concurrent tasks, when it can be done with little or no cognitive effort, and/or when it is not influenced by increases in the sensory information available during performance of the task. Tasks of conscious perception can sometimes meet these criteria for automaticity, either because they involved innately privileged

Future research on attentional interactions involving seeing and acting

Lurking beneath the surface of our normally smooth visual-motor interactions with the environment is the fact that there are two independent streams of visual processing that make use of the same pattern of light. One of these streams enables us to consciously identify objects and to apprehend their layout in space; the other serves to unconsciously facilitate our motor interactions with these objects. Although most previous research on the attentional limitations of vision have focused on the

References (47)

  • G. Binsted et al.

    Visuomotor system uses target features unavailable to conscious awareness

    Proceedings of the National Academy of Sciences of the United States of America

    (2007)
  • B.G. Breitmeyer et al.

    Unconscious color priming occurs at stimulus- not percept-dependent levels of processing

    Psychological Science

    (2004)
  • E. Brenner et al.

    Perceptual requirements for fast manual responses

    Experimental Brain Research

    (2003)
  • R.G. Brown et al.

    An unusual enhancement of motor performance during bimanual movement in Parkinson's disease

    Journal of Neurological Neurosurgery and Psychiatry

    (1998)
  • B.D. Cameron et al.

    The hand's automatic pilot can update visual information while the eye is in motion

    Experimental Brain Research

    (2009)
  • B.D. Cameron et al.

    Dual-target interference for the ‘automatic pilot’ in the dorsal stream

    Experimental Brain Research

    (2007)
  • U. Castiello

    Grasping a fruit: Selection for action

    Journal of Experimental Psychology: Human Perception and Performance

    (1996)
  • U. Castiello et al.

    Temporal dissociation of motor responses and subjective awareness

    Brain

    (1991)
  • R. Chua et al.

    What the hand can't tell the eye: Illusion of space constancy during accurate pointing

    Experimental Brain Research

    (2005)
  • M. Desmurget et al.

    Role of the posterior cortex in updating reaching movements to a visual target

    Nature Neuroscience

    (1999)
  • H. Deubel et al.

    Can man bridge a gap?

    Behavioral and Brain Sciences

    (1994)
  • H. Deubel et al.

    Selective dorsal and ventral processing. Evidence for a common attentional mechanism in reaching and perception

    Visual Cognition

    (1998)
  • J. Diedrichsen et al.

    Independent on-line control of the two hands during bimanual reaching

    European Journal of Neuroscience

    (2004)
  • Cited by (10)

    • Attention is needed for action control: Further evidence from grasping

      2012, Vision Research
      Citation Excerpt :

      This finding is especially interesting as there is an on-going debate on whether dorsal (visuomotor) and ventral (perceptual) processing share common processing resources (Enns & Liu, 2009, chap. 12; Kunde et al., 2007; Liu, Chua, & Enns, 2008; Norman, 2002). Recently, it was suggested that only action planning (which is considered a function of the ventral stream) relies on the same resources as conscious perceptual tasks whereas action control (executed by the dorsal stream) was considered to be unaffected by an attention-demanding ventral stream task (Enns & Liu, 2009, chap. 12). This interpretation was supported by the finding that RTs (indicator for movement planning), but not MTs (indicator for action control) showed significant dual-task costs when a perceptual and a visuomotor task were performed concurrently (Liu, Chua, & Enns, 2008).

    • Delayed pointing movements to masked Müller-Lyer figures are affected by target size but not the illusion

      2011, Neuropsychologia
      Citation Excerpt :

      Based on these results, Franz et al. (2009) argued that the increased effect of illusion after a delay is caused by the availability of visual feedback leading to online corrections of the movement (and not by a shift in control from vision for action to vision for perception). In unrestricted viewing situations, observers cannot perform an action directed at an object without simultaneously perceiving the object (Milner & Goodale, 2008; see also Enns & Liu, 2009; van Doorn, van der Kamp, de Wit, & Savelsbergh, 2009). This makes it difficult to unequivocally determine whether observed illusion effects are a function of vision for perception, vision for action, a combination of both, or of a single all-purpose visual system.

    View all citing articles on Scopus
    View full text