Earlier today we mentioned that last week, Nobel-prize winning scientist Eric Kandel wrote about the five most unforgettable books on memory for The Wall Street Journal. One of the titles in the article was Memory From A to Z by Yadin Dudai. Below is an entry excerpt from the book.
Complex, stereotyped vocalizations, accompanied by characteristic body postures, produced predominantly by mature male birds during the breeding season. Male birds sing to selected audiences. The male is a landlord and potential warrior, notifying other males that it is ready to defend its territory. It is also a charming troubadour attempting to convince females that it is the best in town. The song occupies such a cardinal role in the male’s life that it may even dream about it (Dave and Margoliash 2000). Whereas we humans could enjoy the song repertoire regardless of gender, the male and the female of songbirds are probably each tuned to understand only that part of the song that speaks to their heart (Williams and Nottebohm 1985). The *plasticity of birdsong has been well known to bird fanciers in the Orient since ancient times, and expert manipulations of song were exploited for aesthetic and commercial purposes (Konishi 1985). This neuronal and behavioural plasticity has also long attracted scientists’ attention (Darwin 1871; Mertfessel 1935; Koehler 1951; Thorpe 1954). In addition to being a beautiful system to investigate ethology and learning, the study of birdsong taps into several central issues in brain research. These include the role of genetic constraints on learning (‘prepared learning’, see *a priori, *imprinting); the interplay of *development and learning; the contribution of ‘instruction’ and ‘selection’ processes in learning (see *a priori, *stimulus); and the role of neurogenesis in the adult brain.
A song is a series of sounds with silent intervals between them. It is different from a ‘call’ that is a simple, brief vocalization uttered by both species in all seasons in response to particular stimuli such as a predator. Calls are not unique to birds. Birdsongs are. The most elementary sound in a song is a note, lasting 10–100 ms. Notes form syllables, syllables phrases, and phrases songs. Songs are commonly 1–5 s in duration. Different songs form a repertoire. The size of the repertoire ranges from one to many hundreds songs, depending on the species. Repertoires of geographically distinct populations of the same species often differ, and are termed ‘dialects’ (Baker and Cunningham 1985). Some species perform all or most of their repertoire in cycles that takes many minutes to complete (Marler 1984). Terms such as ‘dialect’ should not lure us to regard birdsong as an analogue of human language, as there is much more to language than structured stereotyped vocalization. But still, the song repertoire provides the bird with a complex expressive and communicative system, which may require special strategies to ensure prompt *retrieval and correct response (e.g.Todt and Hultsch 1998).
The ontogenesis of song involves discrete stages. Take the wild chaffinch as an example (Nottebohm 1970). In the spring, immediately after hatching, chaffinches begin to emit various food-begging calls. Within a few weeks, the male starts to emit a loose, rambling aggregation of low volume notes of varying complexity. This vocal pattern is called ‘subsong’. The subsong keeps changing, and discrete passages, resembling the adult song, gradually emerge. These passages are called ‘plastic song’. During the breeding season the subsong vanishes and the plastic song crystallizes into the full adult song. The singing posture typical of the adult also matures. The final crystallization takes place before the end of the winter.
Although there are remarkable species differences in song development, data from experiments involving sensory and social isolation (e.g. Marler and Tamura 1964; Konishi 1965) generalize to portray the following *model of song ontogeny: the bird is born with a song motor-control system that needs input in order to generate a normal song. This input is provided in two stages, ‘sensory’ and ‘sensorimotor’, which may partially overlap, depending on the species. First comes the ‘sensory stage’, during which the bird listens to a tutor. There is a genetically determined predisposition to prefer a conspecific tutor. Thus, even if we raise a chaffinch in Pavarotti’s house, the chances that it will learn to sing La Bohème are very slim indeed. In the sensory stage, elements of the tutor’s song are confined to memory. In the ‘sensorimotor’ stage, which corresponds to the subsong, plastic song, and crystallization, the bird must listen to itself to match its vocal output with its innate template as well as with the memorized template of the tutor’s song. The entire process combines elements of instruction (by the tutor) and selection (among endogenous innately constrained song templates; see Marler 1997). In the absence of a tutor, only the innate information is used. Some species can generate species-specific song solely on the basis of an innate template, in the absence of tutors and auditory feedback. Species differ also in the stability of song. In some ‘age-limited learners’, such as the zebrafinch and white-crowned sparrow, learning is limited to the first year of life, and crystallized song is maintained throughout adulthood. In ‘open-ended learners’, such as the canary, new songs are added in adulthood. But even adult ‘age-limited learners’ retain a significant amount of plasticity, and use auditory feedback in adulthood to maintain the stability of song structure (Leonardo and Konishi 1999).
One of the advantages of birdsong as an experimental *system is the well defined and quantifiable behavioural output that provides a convenient and faithful *assay to determine whether learning has occurred. Moreover, song is generated by a single organ, the syrinx. This facilitates the tracing of pathways from central motor centres and ultimately identification of brain circuits that subserve *acquisition and execution of the motor programme. Over the years, in series of elegant studies combining anatomical and cellular *methods, a picture has been generated that depicts the major elements of the central song system as composed of two major forebrain pathways (Figure 6). The posteriomedial pathway is traditionally termed ‘the motor pathway’, and includes, in ascending order, the nucleus Uva, the nucleus NIf, the higher vocal centre (HVc, originally so abbreviated because it was thought to be ‘hyperstriatum ventrale’), and finally nucleus RA, that innervates the tracheosyringeal portion of the hypoglossal nerve nucleus (nXIIts), itself innervating the syrinx. This pathway is fed by auditory input, and is obligatory for both song development and production (Nottebohm et al. 1976). Lesions in HVc and RA result in ‘silent song’: upon noticing a female, the lesioned male adopts a singing position but emits no song, becoming a very sad bird indeed. The HVc and RA are organized hierarchically, with HVc neurons representing syllables and RA neurons representing notes. Uva and NIf may help organize syllables into higher units of song. Some of the sites afferent to Uva may also take part in sensory acquisition during song development (Margoliash 1997).
Another interconnected pathway, the anterior forebrain pathway, is considered essential for song development, learning and recognition. It is not obligatory for the mature song production, but still plays a part in feedback evaluation and adaptivity of singing in the adult bird (ibid.; Brainard and Doupe 2000). This pathway indirectly connects HVc to RA via area X, the thalamic nucleus DLM, and the nucleus lMAN. All in all, the song system is distributed over nuclei and circuits, and no single site ‘stores’ the entire score (*engram,*metaphor). Furthermore, a clear-cut dissociation between central ‘sensorimotor’ and ‘learning centres is probably not honoured by the brain.
An intriguing finding is that many new neurons are born in the brain of the adult bird (Goldman and Nottebohm 1983; Alvarez-Buylla and Kirn 1997). Such neurogenesis is not limited to song nuclei, to males, or to species that sing. However, in songbirds, it is prominent in HVc, and correlates with seasonal variations in song and sex hormone levels. (Sex hormones play a part in moulding song circuits and behaviour; Bottjer and Johnson 1997.) The role of neurogenesis in song memory, if at all, is not yet clear. In recent years neurogenesis has also been noted in the adult mammalian brain, and, furthermore, reported to be enhanced in learning (Gould et al. 1999; *hippocampus; but see concerns in Rakic 2002).Neurogenesis in birds in general and songbirds in particular may therefore reflect a more general process. This is surely a finding that can defeat the popular notion that old brains only fade out.