The Binding Problem

Abstract

voice in a crowd; binding across time is required for interpreting object motion; and cross-modal binding is Since its original formulation as a theoretical problem required to associate the sound of a ball striking a bat (von der Malsburg, 1981), "the binding problem" has with the visual percept of it, so that both are effortlessly captured the attention of researchers across many disciperceived as being aspects of a single event. I like to plines, including psychology, neuroscience, computarefer to these sorts of problems as perceptual binding tional modeling, and even philosophy. Despite the isproblems, since they involve unifying aspects of persue's prominence in these fields, what "binding" means cepts. In addition, there are cognitive binding problems: is rarely made explicit. In this paper, I will briefly survey they include relating a concept to a percept, such as the many notions of binding and will introduce some linking the visual representation of an apple to all the issues that will be explored more fully in the reviews semantic knowledge stored about it (it is edible, how it that follow. tastes, used in pies, etc.); cross-modal identification, such as being able to identify an item that has previously What Is Binding?only been seen by how it feels; and memory reconstruc-The canonical example of binding is the one suggested tion, the linking of previously encoded information to by Rosenblatt (1961; see also von der Malsburg, 1999 form a structured and unified representation. While this [this issue of Neuron]), in which one sort of visual feature, perceptual/cognitive distinction is somewhat artificial, such as an object's shape, must be correctly associated it serves to highlight the fact that binding occurs in many with another feature, such as its location, to provide a different kinds of brain processes. The reviews that folunified representation of that object. Such explicit assolow tend to focus on the visual binding problem, alciation, or "binding," becomes especially important though not exclusively. It behooves us to be as explicit when more than one visual object is present, in order as possible about the nature of the particular problem to avoid incorrect combinations of features belonging under investigation, for while it is likely that most differto different objects, otherwise known as "illusory conent forms of binding problems are solved via common junctions" (Triesman and Schmidt, 1982). Considerable mechanisms, that is only a parsimonious assumption, psychological evidence exists for the occurrence of illuand ultimately it is an empirical issue. sory conjunctions (see Wolfe and Cave, 1999 [this is-On the other side of the coin, it is worthwhile to resue]), suggesting that in certain cases, binding is indeed member that something as complex as binding, writ a problem for the brain. In addition, evidence from neularge, may not have a single mechanistic solution. The roanatomy and neurophysiology indicates that propotential mechanisms for binding suggested in these cessing streams in the...

Seeing properties of an invisible object: Feature inheritance and shine-through

“…Feature inheritance and shine-through might be considered as two states of feature binding (23). In feature inheritance only one object, the grating, is perceived and features in a small spatiotemporal window are bound to it.…”

Multi-sensory integration of spatio-temporal segmentation cues: one plus one does not always equal two

“…With a moving object, abrupt changes in the object's speed or direction of movement are essential data for segmentation, either by human observers ), or by non-biological, event-parsing systems (Rui & Anandan, 2000;Agam, Bullock, & Sekuler, 2005). Under many conditions, visual segmentation cues are accompanied by correlated inputs from other senses, notably audition (Roskies, 1999;Shimojo & Shams, 2001). Contentbased, automated image recognition systems can improve their segmentation of events by exploiting correlated multisensory information (Rui, Gupta, & Acero, 2000), but can humans do the same?…”

Emulating the Perceptual System of the Brain for the Purpose of Sensor Fusion

“…Information Integration over Time To perceive objects, events, and situations in an environment, single-moment snapshots of sensory information provided by different modalities is not always sufficient for unambiguous perception. The course and the succession of sensory signals over time is of importance [24].…”