Similarly to "recall, recognition (re-cognoscere, Latin for to get to know again) refers to a type of memory (definition 1, Mandler 1980), a brain process (definition 2), and a memory test (definition 3). To judge that something has occurred previously may mean different things, ranging from the detection of familiarity or recency to the identification of the specific attributes of the target in its proper "context. Imagine entering a classroom and detecting a new student in the front row. Her face looks familiar, but you have no idea who she is. In this case recognition means detection of familiarity only. But now suppose you enter the classroom and realize that the person in the front row is Ann, whom you first met last month at the Faculty club. This is also recognition, but this time, it involves much more knowledge. Many recognition tests in the laboratory measure only familiarity, or recency among equally familiar targets.
The aforementioned types of knowledge are reflected in an influential "model of recognition, called the dual-process model (Juola et al. 1971; Mandler 1980; Jacoby 1991; Yonelinas 1999). As the name implies, this model depicts recognition as being subserved by two qualitatively different processes. One process is the judgement of familiarity; it is generally considered to be automatic and fast.1 The other process is termed recollection, or search, or retrieval.2 It is considered to be intentional and slower, and refers to the retrieval of information about the target and its context. This complements the familiarity decision, and identifies specific attributes of the target and the context. The familiarity process is also labelled K (for 'know') and the recollection process R (for 'remembe/). The terms 'recollection', 'remember', and 'know' connote "conscious awareness, and indeed, the use of these terms in the context of recognition stems from human memory research, where remembering and knowing is "declarative (e.g. Tulving 1985 b). One can, however, adapt the dual process model to animal studies where conscious awareness is not assumed. An interesting spin-off of the dual-process model is the blurring of distinction between recall and recognition. Prominent models of recall propose two stages, the second of which is recognition (see there). A combination of the two-stage model of recall and the dual-process model of recognition, depicts recall as involving a search phase in which the target is generated, followed by familiarity judgement and recollection. But recollection itself could involve activation or reconstruction of "internal representations in the absence of the target, which is bona fide recall. This situation reinforces the notion that both recall and recognition are processes along the continuum of "retrieval (Tulving 1983). These situations differ in the information sought in each case (about the context in recognition, about the target in recall; Hollingworth 1913), and in the retrieval cues (provided by the experimenter or the environment in recognition, self-generated in recall). The clearest distinction between recognition and recall is therefore in the test situation, not necessarily in the type of processes employed by the subject to perform the task.3
Whereas whether animals are capable of genuine recall is open to debate, ample evidence proves that they can surely recognize (e.g. "classical conditioning, "instrumental conditioning, "delay task). Two points are noteworthy. First, is the recognition of only detection of familiarity, or also identification of the specific attributes and context of the target, as we humans recognize? The answer is likely to depend on the species and the task, but in any case requires the proper attention in construing the data. Second, in those tests in which recognition is judged by a decreased response to a familiar stimulus, the possibility that the "performance represents "habituation should not be ignored.4 These caveats notwithstanding, there are clearly tasks in which animals can recognize better than humans, due to superior sensory sensitivity, discriminability, and possibly categorization in the domain of the relevant sensory modality. A beautiful "classic example is provided by Argos, Odyssey's dog: the old dog is the first to recognize Odyssey upon his much belated return to Ithaca, probably by his scent, whereas the nurse Eurykleia needs much more time and the scar on her master's leg as a retrieval cue (Homer, Odyssey, Books XVII, XIX). There are, of course, tasks in which the recognition capacity of the human brain is impressive (Dudai 1997a). For example, in one study subjects learned to recognize no less than 10000 pictures and the author concluded that even a million are possible (Standing 1973).
What are the brain substrates of recognition? This question can be broken into the following subquestions: (a) What is the flowchart diagram of the circuits that subserve recognition? (b) What is the role of each of the stations in theses circuits in the different postulated sub-processes of recognition? and (c) What are the cellular and molecular mechanisms that subserve these roles in each of the stations? The experimental "systems used in this type of research range from recognition tasks, mostly visual, in humans and in the "monkey, to recognition tasks involving the chemical and other senses in rodents (e.g. Mishkin and Murray 1994; Nakamura and Kubota 1996; Schacter et al. 1996c; Reed and Squire 1997; Tanaka 1997; Tulving and Markowitsch 1997; Aguirre and Farah 1998; Berman et al. 1998, 2000; Brown and Xiang 1998; Murray and Mishkin 1998; Parker and Gaffan 1998; Steckler et al. 1998a,b; Suzuki and Eichenbaum 2000; von Zerssen et al. 2001). The "methods involved are selective lesions in experimental animals and analysis of brain damage in "amnesics, cellular physiology and molecular biology in laboratory animals, and in recent years, "functional neuroimaging of human subjects. To this one should add psy-chophysics and modelling, that contribute to the understanding of the "algorithms involved (e.g. Wallis and Bulthoff 1999). But before saying a few words on some general conclusions of this research, it helps to ask what is it that we expect the system to contain. Well, it must contain, as any other memory system, circuits that store the information and are able to retrieve it. But there is an additional component that is a must for recognition: a comparator, that matches the target with stored internal representation, and detects familiarity/novelty. A brainstem-thalamocortical loop was suggested to subserve this function in taste recognition (Berman et al. 2000); other circuits, including corticocortical ones, could fit as well in other systems.
The storage and retrieval circuits that subserve recognition are in general similar to those that subserve recall (Haist et al. 1992; Zola-Morgan and Squire 1993), and include the "limbic archicortex, paleocortex, neocortex, and neuromodulatory systems that regulate "acquisition, "consolidation, and retrieval, including monitoring ("cerebral cortex, "hippocampus, "metamemory, "retrieval). The identity of the areas and their subdivisions and their relative contribution depend on the task type and particulars, e.g. sensory modality involved, or whether the task is spatial or not. Within these systems, there are species-adapted specializations that allow recognition of specific features in the world, such as letters and digits in humans (Polk and Farah 1998), faces in mammals (Gross et al. 1972; Kendrick and Baldwin 1987; Desimone 1991; Golby et al. 2001), or species-specific melodies in birds ("bird-song). Furthermore, different components have been proposed to subserve differentially the two postulated subprocesses of recognition. Hence it was proposed that a neuronal system centred on the perirhinal cortex contributes preferentially to the familiarity judgement, whereas the system centred on the hippocampal formation contributes preferentially to recollection (Brown and Xiang 1998; see also Eldridge et al. 2000). Similarly, there is evidence that in word recognition, the left pre-frontal, left parietal, and posterior cingulate regions contribute to recollection more than to familiarity judgement (Henson et al. 1999).Within cortical circuits that subserve recognition, discrete types of cellular responses are discerned, including a characteristic suppression of the neuronal response to a stimulus once it becomes familiar (Desimone 1996; Brown and Xiang 1998; "priming5). Recognition learning may also take advantage of the presence in the cortex of specific molecular 'novelty switches', which are activated only by unfamiliar but not by familiar stimuli, and whose activation triggers "intracellular signal transduction cascades, culminating in use-dependent long-term "plastic changes in the circuit that ultimately encodes the familiarity (Berman et al. 1998,2000).
Selected associations: Delay task, Habituation, Imprinting, Recall, Surprise
1The detection of familiarity could be examined in the context of system detection theory (e.g. Yonelinas 1999). This theory (Wickens 1984) is applicable to situations in which there are two or more states of the world that cannot be easily discriminated, and whose discrimination involves two response categories, yes or no, e.g. familiar or unfamiliar. This discrimination is not absolute, but rather based on a number of factors, including the discriminability of input signals and the 'criteria used by the 'system, themselves shaped by the price paid for false decisions. These factors depend on experience and 'context. The signal detection theory was originally developed to deal with the detection of sonar and radar signals in the Second World War. It is a prominent example of the application of engineering and information processing theories to human psychology; a related example is the application of such theories to 'attention. 2It is important to note that the use of the term retrieval in the dual-process theory of recognition refers to a limited aspect of retrieval (see also 'recall). Retrieval in general is a universal 'phase of mem-ory,without which no stored information can be actualized. Detection of familiarity, by definition, involves retrieval as well, in this case of the old 'internal representations that are compared with the new ones. It is, however, the convention in the discussion of the dual-process models of recognition to use the term 'retrieval' (or 'recollection', or 'search') to refer specifically to the retrieval of information about the identity of the recognized item. 3Again, the distinction between recognition and recall becomes even fuzzier when recognition pertains to the judgement of previous occurrence of mental images, propositions, and concepts in the train of thought, rather than of sensory 'stimuli only. Hence when one recognizes a thought as contemplated previously, it involves recall and recognition intimately combined.
4The relationship of habituation to recognition is not trivial.A 'reductionist view considers habituation as rudimentary recognition, because the nervous system functions differently if the stimulus has occurred previously. However, the simplest reflex habituation does not involve matching of stored representations with the percept of the on-line stimulus (e.g. 'Aplysia). Complex habituation, such as that of the mammalian orienting reflex ('sensitization), may already involve such matching to detect novelty (Sokolov 1963b).
^Although in the cortex cellular response in recognition may resemble that seen in repetition 'priming, priming is not the source of familiarity in recognition (Stark and Squire 2000).
Was this article helpful?