The magical number 4 in vision
Behavioral and Brain Sciences
Gormican 1988) For instance, the dimension "colour" of an object can be characterized by one of the features "red," "blue," "purple," and so on. After breaking down a visual object into these more basic elements, the question arises as to whether the above definitions of "dimensions" and "features" in vision constitute an appropriate basis for characterizing chunks in visual memory. In other words, one may ask (1) whether chunking in visual working indeed functions by binding features into
... features into integrated visual objects, and (2) analyse the rules and limitations of this binding. With regard to the first question there is recent evidence from Luck and Vogel (1997, also referred to in sect. 3.1.1), suggesting that the capacity limit of visual STM indeed refers to feature bundles in the form of objects. In their experiments, subjects were required to retain simple geometrical visual objects made up of feature conjunctions such as of a certain colour, orientation, and length. The data showed that objects defined by conjunctions of two or more dimensions (e.g., a line of a certain orientation, colour, and length) can be retained as well as objects defined by only a single dimension (e.g., orientation only). For any of these combinations, the estimated memory capacity was about four objects. The second question is strongly related to the problem of how many of these basic elements can be bound into a single chunk, and whether there exist limitations as to the possible combination of features. In our own experiments, we attempted to replicate this surprising finding (Deubel et al., in preparation). We used the same stimuli as Luck and Vogel (1997), and identical experimental parameters such as presentation and retention times. Our experimental results (Fig. 1) clearly show that retention of objects defined by a conjunction of two colours leads to a strong drop in performance, as compared to the condition in which the objects consisted of one colour only. So, external objects with two colours seem to require two chunks for the internal coding. This finding is in obvious contrast to the result of Luck and Vogel (1997). The reason why we could not replicate their data is unclear to us, however, in an independent study, Wheeler and Treisman (submitted) recently reported a finding similar to ours. These data are clear evidence that there exist prominent limitations to chunking in visual memory. As a possible, preliminary rule of thumb suggested by the result, one may assume that a visual chunk can consist of not more than one feature per dimension, that is, one colour, one shape primitive, and so on. A further, yet unresolved important issue in this context is the question whether there is also a limit in the number of possible dimensions that define a chunk. Luck and Vogel (1997) found no limit (i.e., no drop in memory performance) up to a conjunction of four different dimensions (colour, orientation, length, gap). However, it might be that a limitation larger than that can indeed be found. The empirical task of the future will be to determine more precisely the limitations of chunking and how they relate to visual features and dimensions. Indeed, we think that the paradigm presented here offers a promising experimental approach to answer questions about the nature of chunks in vision. Measuring memory performance for a variety of stimuli and features could reveal the basic dimensions and features in vision in a very straightforward way: If adding the feature in question to the stimuli leaves the memory capacity (in terms of number of objects) unaffected, one may conclude that it is really a basic visual feature, forming an elementary part of a visual chunk. Cowan defines a chunk as "a collection of concepts that have strong associations to one another and much weaker associations to other chunks currently in use." This definition does not impose any constraints on the nature and number of elements that can be bound into a chunk. Our experiment is a demonstration that such limitations exist, and that their analysis may lead to important insights into properties of visual memory.