Josep B. Trobalón

Statistical computations over a speech stream in a rodent

Statistical learning is one of the key mechanisms available to human infants and adults when they face the problems of segmenting a speech stream (Saffran, Aslin, & Newport, 1996) and extracting long-distance regularities (Gómez, 2002; Peña, Bonatti, Nespor, & Mehler, 2002). In the present study, we explore statistical learning abilities in rats in the context of speech segmentation experiments. In a series of five experiments, we address whether rats can compute the necessary statistics to be able to segment synthesized speech streams and detect regularities associated with grammatical structures. Our results demonstrate that rats can segment the streams using the frequency of co-occurrence (not transitional probabilities, as human infants do) among items, showing that some basic statistical learning mechanism generalizes over nonprimate species. Nevertheless, rats did not differentiate among test items when the stream was organized over more complex regularities that involved nonadjacent elements and abstract grammar-like rules.

The use of prosodic cues in language discrimination tasks by rats.

Recent research with cotton-top tamarin monkeys has revealed language discrimination abilities similar to those found in human infants, demonstrating that these perceptual abilities are not unique to humans but are also present in non-human primates. Specifically, tamarins could discriminate forward but not backward sentences of Dutch from Japanese, using both natural and synthesized utterances. The present study was designed as a conceptual replication of the work on tamarins. Results show that rats trained in a discrimination learning task readily discriminate forward, but not backward sentences of Dutch from Japanese; the results are particularly robust for synthetic utterances, a pattern that shows greater parallels with newborns than with tamarins. Our results extend the claims made in the research with tamarins that the capacity to discriminate languages from different rhythmic classes depends on general perceptual abilities that evolved at least as far back as the rodents.

Locating an invisible goal in a water maze requires at least two landmarks

In three experiments, rats were trained in a spatial task in a swimming pool similar to that used by Morris (1981), but surrounded by black curtains that provided a great level of control over the landmarks that defined the location of the goal. Experiment 1 showed that rats can learn to find an invisible platform in a fixed location relative to four landmarks in a very homogeneous environment. Experiments 2 and 3 demonstrated that the animals need two landmarks to recover the information about the environment stored in their memory as a cognitive map.

Perception of two vowel contrast by rats: discrimination of natural tokens

La percepcion fonetica ha sido estudiada en diferentes especies animales usando silabas aisladas, vocales o consonantes, proporcionando varios resultados similares entre humanos y animales. Sin embargo, la medida en que estas similitudes se encuentran en el procesamiento de estimulos naturales mas complejos es todavia un tenia de debate. En este manuscrito presentamos dos experimentos sobre la percepcion de distintos contrastes vocalicos en ratas. Se usaron tres pseudo palabras, producidas por varias locutoras, que contenian el contraste a estudio. En el Experimento 1 las ratas pudieron diferenciar diferentes ejemplares de la pseudo palabra [dedil de ejemplares de la pseudo palabra [dudi/. En el experimento 2, esta discriminacion fue tambien observada en pseudo palabras que contenian vocales mas cercanas: [dodi]y [dudi]. Los resultados muestran que las ratas pueden normalizar a traves de diferentes caracteristicas del habla natural, y que son capaces de percibir contrastes vocalicos dentro de palabras.

Discriminación de frases y generalización en ratas

espanolThe ability to detect features in speech signals that permit language discrimination is present early in life in human newborns. It has been suggested that this ability may help the child to bootstrap linguistic parameters and thus contribute to his/her language development. Additionally, the study by Ramus, Hauser, Miller, Morris & Mehler (2000) with cotton-top tamarin monkeys, has shown that this ability may also be present in other species. This finding was corroborated by Toro, Trobalon & Sebastian-Galles (2003), who observed that, after training, rats were also able to discriminate between synthesized Dutch and Japanese sentences, but could not do so when the sentences were played backwards. The observation of this pattern in human newborns, tamarin monkeys and rats, supports the claim that the ability may be linked to other functions during the analysis of the acoustic signal. To explore whether a single sentence contained sufficient information for rats to identify features that differentiate between two languages, and whether they could generalize these features to new sentences, we performed two separate tests of their ability to discriminate between sentences in Dutch and Japanese. The first test showed that rats were able to learn to distinguish between two single sentences. The second demonstrated that they could extract relevant cues from a single sentence, and apply subsequently these cues to two new synthesized sentences of these languages EnglishLa habilidad para detectar claves relevantes en el habla que permitan la discriminacion entre lenguas esta presente desde temprano en los bebes. Se ha sugerido que esta habilidad puede ayudar al nino en el arranque (bootstrap) de parametros linguisticos y por lo tanto, contribuir al desarrollo del lenguaje. Sin embargo, el trabajo de Ramus, Hauser, Miller, Morris y Mehler (2000) con monos tamarines, ha mostrado que esta habilidad puede estar presente en otras especies, tal como fue luego demostrado por Toro, Trobalon y Sebastian-Galles (2003). Luego de recibir entrenamiento, las ratas tambien podian discriminar entre frases sintetizadas de Holandes y Japones, mientras que no lo podian hacer cuando estas frases eran presentadas hacia atras (backwards), un patron que se ha observado tambien en bebes humanos y monos tamarines, lo que apoya la afirmacion de que esta habilidad puede estar relacionada con otras funciones al analizar las senales acusticas. Para investigar si solo una frase contiene la informacion necesaria para que las ratas puedan encontrar las caracteristicas diferenciadoras entre dos lenguas, y si pueden generalizar estas a nuevas frases, llevamos a cabo dos pruebas sobre las habilidades de discriminacion de las ratas de frases de Holandes y Japones. La primera prueba mostro que las ratas si aprendieron la discriminacion. La segunda demostro que las ratas pueden extraer las claves relevantes y generalizarlas a dos nuevas frases sintetizadas de estas lenguas