Types of attention matter for awareness: A study with color afterimages

Abstract

It has been argued that attention and awareness might oppose each other given that attending to an adapting stimulus weakens its afterimage. We argue instead that the type of attention guided by spatial extent and perceptual levels is critical and might result in differences in awareness using afterimages. Participants performed a central task with small, large, local, or global letters and a blue square as an adapting stimulus in three experiments and indicated the onset and offset of the afterimage. We found that increases in the spatial spread of attention resulted in the decrease of afterimage duration. In terms of levels of processing, global processing produced larger afterimage durations with stimuli controlled for spatial extent. The results suggest that focused or distributed attention produce different effects on awareness, possibly through their differential interactions with polarity dependent and independent processes involved in the formation of color afterimages.

Introduction

Attention is closely related to awareness and some have argued that it is necessary for awareness (Dehaene et al., 2006, Mack and Rock, 2001, Posner, 1994). Studies using various paradigms such as inattentional blindness and change blindness have shown that attention not only affects conscious identification of stimuli (Lavie, 2006, Mack and Rock, 2001, Rensink, 2002) but also their phenomenal appearance (Carrasco, Ling, & Read, 2004). On the contrary, other studies have questioned the necessity of attention for awareness (Koch and Tsuchiya, 2007, Lamme, 2003, Tsuchiya and Koch, 2005).

According to Lamme (2003), attention operates on conscious stimuli leading to verbal report rather than consciousness resulting from attended stimuli. Koch and Tsuchiya (2007) proposed a fourfold classification scheme of aware and unaware percepts and behavior into four psychophysically defined categories depending on whether top-down attention is necessary and whether it gives rise to awareness. The organization of these categories include situations and tasks in which (i) attention gives rise to awareness (e.g. detection, discrimination and reportability), (ii) inattention does not give rise to awareness (e.g. rapid vision), (iii) attention does not give rise to awareness (e.g. priming, visual search) and (iv) attention is absent while awareness is present. The fourth case of processes that do not need top-down attention but gives rise to consciousness is based on findings of better performance under conditions of inattention (Kanai & Verstraten, 2006; Leopold, Wilke, Maier, & Logothetis, 2002; Li et al., 2002, Lou, 2001, Suzuki and Grabowecky, 2003).

Kanai and Verstraten (2006) asked the observers to indicate the direction of motion of a bistable stimulus using ambiguous motion. They showed that the priming effect of an unambiguous motion stimulus reduced when attention was drawn by a secondary task of letter identification performed during the interval between adaptation with the prime stimulus and the ambiguous motion stimulus (Kanai & Verstraten, 2006). Li et al. (2002) used a dual task paradigm to evaluate the role of focused attention when observers performed a categorization task with objects present in natural scenes at the periphery like animal vs. non-animals and vehicle vs. non-vehicles. The observers showed good performance on the peripheral scene-categorization task with a relatively difficult central visual search task in which observers had to search for an odd element in an array of five randomly rotated Ls or Ts. However, when a letter discrimination task was performed with the same central task, the observers performed at chance level. The study indicated that certain properties of natural scene such as their gist could be accessed with no or very little focused attention (Li et al., 2002).

More pertinent to the current study is the set of findings from studies involving attention and color afterimages (Lou, 1999, Lou, 2001, Suzuki and Grabowecky, 2003, Tsuchiya and Koch, 2005, Wede and Francis, 2007a) that have been used to argue for the opposing effects of attention on awareness (Koch & Tsuchiya, 2007). Suzuki and Grabowecky (2003) measured afterimage durations and their time of onset/offset in a series of experiments in which participants were asked to focus their attention to one of two overlapping triangle inducers. The afterimage for the unattended triangle appeared earlier and disappeared later compared to the afterimage of the attended triangle. In another experiment by them, the participants either attended to the afterimage inducer or to a stream of letters. They found that afterimage onset was delayed when the afterimage inducer is attended compared to when the letter stream was attended or inducer is not attended (Suzuki & Grabowecky, 2003).

In yet another study (Lou, 1999), the participants were presented with two overlapping inducer triangles that produced color afterimages of the two triangles. When attention was directed to one of the two afterimages, the attended afterimage disappeared faster than the unattended afterimage (Lou, 1999). These results have been used to support the argument that attention and awareness can have opposing effects on each other (Koch & Tsuchiya, 2007).

Attention is not a unitary process and different types of attention might have different relationships to awareness. One way to characterize attention would be in terms of focused attention and distributed attention (Chong and Treisman, 2005, Demeyere and Humphreys, 2007, Treisman, 2006). Distributed attention involves processing at larger spatial scales whereas focused attention is typically associated with processing at smaller spatial scales (Treisman, 2006). The distributed attention mechanisms involve parallel processing that contrasts with the focused attention mode that individuates objects (Treisman, 2006). Chong and Treisman (2005) showed that that computation of statistical properties of a visual scene such as mean size is compatible with conditions when attention is distributed globally compared to when attention is focused on individual items. Although automaticity accounts have been proposed for explaining distributed attention (Treisman, 2006), others have shown that extraction of statistical properties may be rapid but not automatic (de Fockert & Marchant, 2008).

It is possible that the distributed attention mechanisms are recruited when focused attention is not an optimal strategy for perception. For example, even when poor identification of individual items such as orientation signals is difficult (in crowded displays), the visual system accurately estimates the average tilt (Parkes, Lund, Angelucci, Solomon, & Morgan, 2001). Evidence for relative independence of focused and distributed attention mechanisms comes from studies with simultanagnosic patients who were shown to be sensitive to different forms of information (e.g. color, size) when in distributed attention mode rather than focused attention mode (Demeyere and Humphreys, 2007, Demeyere et al., 2008). Despite impairment in simultaneously encoding information from a visual display, which manifests as an inability to count more than one or two items, the simultanagnosics were able to estimate the number of stimuli with accuracy better than chance level (Demeyere & Humphreys, 2007).

Given differences in the mechanisms associated with focused vs. distributed attention, they might be linked to different types of awareness. For example, Block (2005) proposed a framework that distinguishes phenomenal and access awareness. Phenomenal awareness refers to experiential aspects of perception while access consciousness refers to conscious content that is broadcast to many systems in the brain and is closely linked to cognitive processes like executive attention, planning, and voluntary control. Whereas phenomenal awareness can be instantiated by brief parallel processing of objects, access awareness necessitates zooming into individual objects so that they can be eventually acted upon (Block, 2005). There is a potential possibility of phenomenal awareness and access awareness being facilitated by mechanisms of distributed attention and focused attention, respectively. In a related vein, object consciousness could be associated with focused attention and background consciousness with distributed attention (Iwasaki, 1993).

The goal of the present study was to investigate differences in awareness due to focused and distributed attention rather than attention vs. inattention. We focus on the findings from afterimages to understand better the relationship between different types of attention and awareness. We manipulated the spatial spread of attention (attentional spotlight) and level (global or local) of attention using a central task during the adaptation period and measured its effect on the adapting stimulus. Focused or distributed attention was manipulated by using letter stimuli of different sizes and hierarchical (global and local) stimuli. For example, attending the small sized single letters (Fig. 1a) would involve focused attention to a small spatial scale. Whereas, attending to large sized letters would involve either focused (Fig. 1b) or distributed (Fig. 1d) attention to large spatial scales.

Hierarchically structured stimuli provide a unique way to study such distributed attention mechanisms given its focus on the “forest” or “whole” rather than the “trees” or “parts”. Hierarchical stimuli typically consist of a large shape (global) made up of smaller shapes (local). Both global and local levels are processed in parallel at early stages of cortical processing (Heinze & Münte, 1993) and their processing diverges at higher stages of perceptual processing (Heinze, Hinrichs, Scholz, Burchert, & Mangun, 1998). The differences in global and local levels of processing especially with hierarchical stimuli are partly based on spatial frequency processing (Fink et al., 1999, Sasaki et al., 2001, Shulman and Wilson, 1987) and spatial tuning, with global attention to the large shape being associated with enhancement over a larger spatial scale compared to local attention (Sasaki et al., 2001). Yet, local attention to small shapes that make up a large shape may be associated with a larger spatial extent compared to focused attention to an individual element. This may be possible due to grouping of local elements contributing to greater spatial spread during local processing (Han et al., 2002). Apparently, a stimulus with a smaller size would involve attending to a smaller region compared to the same stimulus with larger size. Even with a spatial region selected for further processing, attention could be manipulated in terms of different levels of object-based processing.

The role of object-based representations for awareness has been emphasized in a binocular rivalry study where the cued surface dominated subsequent perceptual rivalry even when the dominant and the suppressed surfaces occupied the same region of space (Mitchell, Stoner, & Reynolds, 2004). This indicates that processes involved in awareness are affected by the way attention operates on object-based representations. The differences in the level of processing on which attention operates might be important in determining the nature of adaptation aftereffects. Therefore, we were particularly interested in examining the role of focused/distributed attention in terms of spatial spread and levels of object-based processing on awareness through afterimages.

Given the links between attention and awareness, we hypothesized that the type of attention will differentially affect the duration of afterimages. In terms of spatial attention, as attention is more focused with the central task, afterimage durations should be longer with the small compared to the large letter and local compared to global letter. This is based on the findings of stronger afterimages observed when attentional focus was maintained at the centre of the display compared to when attentional scope was broad by performing a task with the peripheral inducer (Suzuki & Grabowecky, 2003).

There is some indication that attentional selection at the global or local level is linked to differences in neural activity associated with awareness (Koivisto, Revonsuo, & Lehtonen, 2005). Koivisto et al. (2005) found interactions between awareness and the type of attention in the later portions of the early anterior negativity (P2-N2 window), with larger differences between recognized and unrecognized stimuli for local attention compared to global attention in the right hemisphere. This interaction indicates that awareness could be influenced by attention to local and global levels. Given potential differences in awareness for the global attention and its link to distributed attention that operates on large spatial scales (Chong and Treisman, 2005, Treisman, 2006), we hypothesized that global target identification would produce stronger afterimages compared to identification of the same-sized single large letter that relies more on focused attention.