Boolean map theory
Boolean map theory
The Boolean Map theory emphasizes the distinction between selection and access in visual attention, and argues that the limits of visual access (i.e., what can be perceived at one moment) boil down to a surprisingly simple data format: a Boolean map. A Boolean map is the linkage of a single feature value per dimension associated with a map (i.e., a set of locations). The Boolean map theory is the driving force behind the studies listed below.
Relevant articles:
You can also download a zip file including all 13 papers that are related to the Boolean map theory.
Visual access of multiple features and locations
A central claim of the Boolean map theory is the distinction between multiple locations and single feature value. In this figure, one can easily verify that the balls on the left side and those on the right side occupy the same set of locations. However, to verify that they include the same set of colors, one would need to check the colors one by one.
Relevant article: 2010, JOV.
Boolean map vs. Object as the Unit of access/ VWM
What are the unit of conscious access (i.e., what can be consciously perceived at one moment) and visual working memory? Here, I challenge the “object-based” account and argue that the unit of both is better defined by the concept of a “Boolean map.” For example, in this figure, there are two dumbbell-shaped objects each consisting of a red and a green end. Various experimental results indicated that we see/remember two red balls (or two green balls) together as a unit, but do not see/remember the information in the dumbbell-shaped object together as a unit.
Relevant articles:
Orientation advantage in VWM
Boolean map theory’s notion of simultaneous representation of multiple locations made a unique prediction. For features, bar orientations, but not colors, can be represented in an all-stimulus-item Boolean map. For tasks, change detection, but not visual search, can rely on the all-stimulus-item Boolean map. Together, a unique advantage for bar orientation in change detection is expected and this is confirmed experimentally. This is the link between the Boolean map theory and FVS framework.
Relevant article: 2015, Psych-Sci
Chunking in working memory via content-free labels
Previous studies found that VWM was enhanced when pairs of colors were predictably paired. We (Huang & Awh) propose that familiar ensembles of items may be attached to a content-free label that can be used to retrieve the constituents of a chunk when they are needed (i.e., panel c). These content-free labels will lead to better performance without real increase in the number of individuated representations stored online. If accessing “compressed” memories requires an additional retrieval process, then access to compressed items should be slower than for uncompressed items. This RT cost was confirmed in a set of experiments.
Relevant article: 2018, Sci-Rep
Attention and figure-ground segmentation
In the left panel, attending to red (or green) regions tends to make those regions appear to be in front of other regions (i.e., become the figure). However, in the right panel, attention has the opposite effect. For example, attending to the red regions makes them appear to be a red face behind green pillars.
Relevant article: 2009, Psych-Sci
Statistical properties demand as much attention as object features
Previous studies have argued that the statistical properties can be extracted with a reduced amount of attention. I showed that the statistical properties demand as much attention as object features (left panel), and it is not the case that statistical properties can be extracted in a cost-free manner (right panel).
Relevant article: 2015, Plos one
Grouping by similarity is mediated by feature selection
A recent account suggests that grouping by similarity is mediated by the attentional selection of a feature. This account makes the distinctive prediction that for grouping by similarity (left panel), but not for low-level groupings (e.g., connectedness, common region, right panel), when two conflicting grouping cues are present, the grouping structure will tend to be driven either by one type of cue or by the other type of cue and will usually not lead to the impression of no grouping. This predicted pattern of results was confirmed.
Relevant article: 2015, PBR
Across-dimension selection is not implemented by a salience map, but by subset selections
Observers saw two adjacent displays and indicated whether the to-be-selected items within the two displays matched in terms of their spatial structure. The observers in our study could readily perceive conjunctively defined subsets (left panel), but had great difficulty with disjunctively defined subsets (right panel). The results pose a challenge to the popular idea that attention is guided by a priority (or salience) map that sums all factors into one single dimension.
Relevant article: 2012, JEP:HPP
Selection is fast but attentional advantage is slow.
In Boolean map theory, we distinguish between 2 concepts that are often conflated with the term attention, namely the selection of information from the stimuli and the following processing optimization (i.e., attentional advantage) of the selected stimulus. I measured the speed of feature-based attention in relation to both definitions. Attention is slow (> 100 ms, panel a) in terms of attentional advantage (e.g., seeing the red digit) but is very fast (< 50 ms, panel b) in terms of the selection of information (e.g., seeing the red pattern). These results support the view that feature-based attention works by creating a spatial representation (i.e., a Boolean map) for the stimulus of a feature.
Relevant article: 2010, JEP:HPP
Familiarity does not aid access to features
Previous findings have suggested that a familiar pattern and the features within it are perceived better than an unfamiliar pattern. I found that comparison of colors in Stars and Stripes flags was equally efficient in the upright and the inverted orientations (top panel). However, the selection of upright Stars and Stripes flags was significantly more efficient than that of inverted flags (bottom panel). I argue that familiarity aids the perception of a pattern only by giving the whole pattern a content-free label and does not directly aid access to features.
Relevant article: 2011, PBR