Generalization decrement and not overshadowing by associative competition among pairs of landmarks in a navigation task.

Abstract

When they are trained in a Morris water maze to find a hidden platform, whose location is defined by a number of equally spaced visual landmarks round the circumference of the pool, rats are equally able to find the platform when tested with any two of the landmarks (Prados, & Trobalon, 1998; Rodrigo, Chamizo, McLaren, & Mackintosh, 1997). This suggests that none of the landmarks was completely overshadowed by any of the others. In Experiment 1 one pair of groups was trained with four equally salient visual landmarks spaced at equal intervals around the edge of the pool, while a second pair was trained with two landmarks only, either relatively close to or far from the hidden platform. After extensive training, both male and female rats showed a reciprocal overshadowing effect: on a test with two landmarks only (either close to or far from the platform), rats trained with four landmarks spent less time in the platform quadrant than those trained with only two. Experiment 2 showed that animals trained with two landmarks and then tested with four also performed worse on test than those trained and tested with two landmarks only. This suggests that generalization decrement, rather than associative competition, provides a sufficient explanation for the overshadowing observed in Experiment 1. Experiment 3 provided a within-experiment replication of the results of Experiments 1 and 2. Finally, Experiment 4 showed that rats trained with a configuration of two landmarks learn their identity.

The effect of disrupting configural information on rats’ performance in the Morris water maze

“…Studies in the Morris water maze have shown that animals trained with four (i.e., A, B, C and D) landmarks performed less accurately when tested with sets of two landmarks alone than animals initially trained with these two landmarks in isolation. This could be because B and C, or D and A landmarks alone are perceived as different from A, B, C and D all together, and the response established to one stimulus configuration cannot be transferred perfectly to a different configuration, resulting in generalization decrement (Pearce, 1987(Pearce, , 1994Chamizo, Rodríguez, Espinet, & Mackintosh, 2012). Another way of expressing this result would be to say that the four landmark case suffers from greater overshadowing of one landmark by the others than the two landmark case, but we note that Chamizo et al (2012) demonstrated that the addition of two new landmarks, and the removal of two old ones, both disrupted performance.…”

“…This could be because B and C, or D and A landmarks alone are perceived as different from A, B, C and D all together, and the response established to one stimulus configuration cannot be transferred perfectly to a different configuration, resulting in generalization decrement (Pearce, 1987(Pearce, , 1994Chamizo, Rodríguez, Espinet, & Mackintosh, 2012). Another way of expressing this result would be to say that the four landmark case suffers from greater overshadowing of one landmark by the others than the two landmark case, but we note that Chamizo et al (2012) demonstrated that the addition of two new landmarks, and the removal of two old ones, both disrupted performance. They argued that these results were consistent with the proposition that a change in the stimulus conditions from the training phase to the test phase led to generalization decrement.…”

“…In the same study, Chamizo et al (2012) investigated whether rats learn about the identity of the landmarks in these experiments. Their results showed that rats knew about the identity of the landmarks learnt during the training phase, because a rat's performance was significantly disrupted by swapping the landmarks original positions.…”

Bloqueo y ensombrecimiento en un grupo de vertebrados filogenéticamente antiguo: los anfibios

“…Pero al mismo tiempo, el ensombrecimiento podría ser susceptible a otras interpretaciones alternativas. Por ejemplo, este fenómeno podría ser explicado mediante un mecanismo de decremento de la generalización: un animal entrenado con un compuesto AB y testeado sólo con B experimenta un gran cambio desde la etapa de entrenamiento a la etapa de prueba, comparado con otro animal entrenado y testeado sólo con B (Chamizo et al, 2012). Para corroborar si éste es el caso para los resultados obtenidos en anfibios, debemos complementar esta información con futuros experimentos en los cuales se incluyan (entre otros elementos) varios lugares de partida dentro del campo abierto.…”

“…En particular, en el diseño implementado en nuestro estudio, una clave visual (clave cercana) provee claramente una mejor predicción de la pileta reforzada que otra (clave lejana), lo cual predice una competencia asociativa en la que el mejor predictor ensombrecerá al peor. Sin embargo, si distintas claves visuales son distribuidas equidistantes alrededor del perímetro del campo abierto y los animales son libres de aproximarse a la meta desde cualquier dirección (en lugar de partir siempre desde el centro de la arena), sería igualmente razonable sugerir que después de una buena cantidad de ensayos todas las claves sean más o menos equivalentes como predictores de la ubicación de la meta, y que no haya necesariamente competencia entre ellas (Chamizo et al, 2012). Esta posibilidad alternativa queda pendiente de ser estudiada más adelante en anfibios.…”