Although some studies have dealt with generalization and discrimination processes in the auto shaping situation, few of them have directed their attention exclusively to the temporal dimensions of the autotyped stimulus. To tackle this study used a group of five pigeons which had been used in a previous experiment with delayed auto shaping. The  CS consisted of the illumination with a red light of a response key for 32 sects.. and that was followed by the presentation of food in 50 % of the trials. After extended exposure to the former training, subjects showed differentiated behavior throughout the presentation of the CS: progressive acceleration in the response rate of auto pecking as the end of the trial and the moment in which the food appeared drew nearer. This pattern of behavior was taken as a baseline for the temporal generalization test that was carried out in the present research.  The temporal generalization study was conducted by installing a peak procedure Catania, 1970; }Roberts. 1981) where in each session, besides the trials with the autoshaped CS of 32 s reinforced in 50% of occasions (food trials), some trials were inserted which were never followed by food and in which the responses key red light remained on 64 s (empty trials). From the autopecking responses of these last trials, a temporal generalization gradient could be observed where, in the steady state, the peak rate approximately coincided with the reinforced duration of 32 s, in spite of there being presented  an area shift of the gradient toward the second half of the empty trials. However, during the first test trials, besides a clear area shift, there was also a lack of adjustment  between the peak rate of responses and the reinforced duration of the stimulus: a clear peak shift of responses due to a significant increase in the rates of responses beyond the 32nd second was appeared. With the prolongation of the differential training with both food trials of 32 s and empty trials of 64s, the peak of responses adjusted progressively to the reinforced duration and, also decreased the magnitude of the area shift.  The data show the course of temporal discrimination and generalization when autoshaped stimuli with a longer duration than the standard ones are used. The results and the technique of peak procedure itself are discussed in the stimulus control framework, defending the suitability of a classic theory such as that by Spence (1937) for the interpretation of the control of behavior by the temporal dimensions of long lasting autoshaped stimuli.Se utilizó un grupo de cinco palomas que, procedentes de una investigación previa, poseían un extenso entrenamiento con un EC automoldeado de 32 s seguido de comida en el 50% de los ensayos. A lo largo del EC los sujetos manifestaron Lll1comportamiento diferenciado consistente en una progresiva aceleración de la tasa de picoteo conforme se acercaba la terminación del EC. Este comportamiento se tomó como línea base para el estudio de la generalización temporal mediante [a instauración de un procedimiento del vértice: ensayos con el EC de 32 s reforzados al 50% (ensayos base) y presentación ocasional de ensayos en extinción con la tecla iluminada durante 64 s (ensayos de prueba). El registro y análisis de la respuesta de picoteo en los ensayos de prueba permitió observar un gradiente de generalización temporal que, en los primeros ensayos, manifestaba desplazamientos del vértice y del área más allá de la duración reforzada de 32 s. La prolongación del entrenamiento diferencial entre ensayos base y ensayos de prueba condujo al progresivo ajuste del vértice de respuestas a la duración reforzada y a la disminución del desplazamiento del área. Los resultados se discuten e interpretan en el marco de la teoría clásica de interacción de gradientes (Spence, 1937).