Four marmosets (one normal male, one blind male, one normal female, and one blind female), with an average age of eight years and average weight of 354 g, were housed in individual cages in a room with attenuated noise, controlled temperature (average temperature of 25.5°C), water at libitum and food daily available for 8.5 hours. The animals were first exposed to an LD cycle of 24 hours (LD 12:12). Illuminance was 150 lux during the light phase and 1 lux during the dark phase. The marmosets were adapted to the laboratory for 10 days. After two weeks in LD 12:12, the time of lights-on was delayed by 6 hours; three weeks later, it was advanced by 6 hours. Four weeks later, the marmosets were placed in constant light conditions for 4 weeks. The marmosets were then returned to LD 12:12 for 3 more weeks and then again to constant light for 4 weeks.
General circadian locomotor activity of the marmosets was measured using an infrared motion sensor above the cage. Output from the sensors was integrated with an IBM-compatible computer running data acquisition software. Analyses of rhythm characteristics and graphical output, actograms, were undertaken using the El Temps computer program (Diez-Noguera, Barcelona, Spain). The free-running period of the locomotor activity rhythm under constant light was computed by the chi square periodogram procedure  with a global risk level (α) of p < 0.05. Under LL, the onset of activity, designated as circadian time (CT) 0, was used as the phase reference point for the onset of the subjective day. Phase-shifts were determined as the difference between projected times of activity onset on the day after dark stimulation. The dark stimulation consisted of 2-hour pulses of darkness. Experiments were in compliance with the institutional guidelines of the Universidade Federal do Rio Grande do Norte and Sociedade Brasileira de Neurociência e Comportamento.