Experimental animals
The experimental animals were a male harbour seal called `Luca` (13 years old at the beginning of experiments) and a male South African fur seal called `Fin` (5 years old at the beginning of experiments). Both animals were kept at the “Marine Science Center” of the University of Rostock, Germany, where they received approximately 50 % of their daily diet during the experimental sessions once or twice per day, five–seven days a week; the rest of the food was provided in training besides the experiments. Both seals were experienced in conducting visual experiments. However the harbour seal had already participated in numerous visual and visual cognitive experiments including the first timing experiment (Heinrich et al. 2016; Scholtyssek et al. 2008; Scholtyssek et al. 2013); whereas, the fur seal’s previous experience was limited to a brightness discrimination study (Scholtyssek and Dehnhardt 2013).
Apparatus
All experiments were conducted in an experimental chamber (3 m deep, 2 m wide, and 2 m high) to achieve a constant illumination via a fluorescent lamp (Standard FSL T8 36W 765 Radium, Wipperfürth, Germany). The illumination, measured with a luxmeter (Voltcraft VC 4 in 1, Multifunctional Environment Measuring Instrument, Conrad electronics AG, Wollerau, Switzerland), was 40 lx in the surrounding of the station of the harbour seal (a drawing of the harbour seal's station can be found in Heinrich et al. 2016). With the fur seal, we first worked with open chamber due to issues related to the motivation of this less experimentally experienced animal; thus, the illumination first was 240 lx at the station of the fur seal. Afterwards, we could continue with closed chamber which reduced the illumination to 60 lx (Fig. 1). The difference in illumination at the stations even in the closed chamber resulted from the different height of the animals’ experimental stations which was varied to enable both animals a natural body position during the experiment. For the harbour seal, the station consisted of a metal hoop affixed to a steel plate 6 cm above the bottom. For the fur seal, the station consisted of a jaw station attached to aluminium profiles (60 cm long, 5 cm wide) at a height of 68 cm (Fig. 1). The experimental stations ensured a constant distance of 50 cm to, and a constant viewing angle of the seals on the LCD monitor (Eizo, Flex scan S1721, 17”, refresh rate 60 Hz, Eizo Nanao Corporation, Hakusan, Ishikawa, Japan), on which the stimuli were presented (see “Stimuli”). Two response targets were fixed on both sides of the respective station devices. The animals moved their heads to one of these response targets after stimulus presentation (see “Procedure”).
During the experiment, the experimenter was hiding in an observation room adjacent to the experimental chamber. Both chambers, the experimental chamber and the observation room, were linked by a window. During each trial, the window was closed by a black opaque slider to prevent secondary cueing. It was opened for rewarding the animal after a correct answer. A camera (HD Logitech Webcam C270, Logitech GmbH, Munich, Germany), installed in the experimental room, allowed to observe the response behaviour of the animals during the session. All technical equipment was located in the observation room allowing the experiment to be operated from this room.
Stimuli
The stimuli, displayed on the LCD monitor, were white-filled circles, 10.5 cm in diameter, on black background. The circle was presented either for the duration of a standard time interval (STI) or for a longer comparison time interval (CTI). The fur seal was tested with eight STIs: 0.2 s, 0.4 s, 0.8 s, 1.6 s, 3 s, 5 s, 7 s, and 12 s. The harbour seal was tested with five STIs: 0.1 s, 0.2 s, 0.4 s, 0.8 s, and 1.6 s; his thresholds for STIs ranging from 3 s to 30 s were already published (Heinrich et al. 2016). This list of STIs illustrates that, in the fur seal, difference thresholds could unfortunately neither be assessed for a 0.1 s STI nor for STIs longer than 12 s as the animal refused to work under these conditions; this generally very unsteadily cooperating animal was unmotivated to keep attention to the very short STIs and to wait for the end of stimulus presentation for long STIs.
The stimuli ranging from 0.2 s to 1.6 s were programmed and presented with PsychoPy 1.82.01 (Peirce et al. 2019; Peirce 2007; Peirce 2009) and the stimuli ranging from 3 s to 12 s were programmed and presented with Matlab R2012b (The Mathworks, Natick, Massachusetts, USA) and the Psychophysics toolbox 3.0 (Brainard 1997; Kleiner et al. 2007; Pelli 1997). We used Matlab to assess difference thresholds in the fur seal for STIs 3 s and longer allowing direct comparison with the harbour seal data from Heinrich et al. (2016). Psychopy was used for STIs 1.6 s and shorter as this programme was more accurate for these time intervals. The accuracy of stimulus presentation was measured with a CMOS camera (Phantom V12, Vision Research Wayne, NJ, USA) recording at 1,000 frames per second. Two human observers counted the number of frames between the frame on which the stimulus had just appeared and the first frame on which it had just disappeared. From the number of frames, the duration of the respective time interval could be calculated and compared to the programmed value. The duration of the time intervals programmed in Matlab deviated by on average 40.0 ± 8.9 ms from the programmed value, and the duration of the time intervals programmed in Psychopy deviated by on average 25 ± 4.7 ms from the programmed value.
Procedure
Each trial started with the experimental animal stationing in its designated station in the experimental chamber. After guiding the animal to its station, the experimenter left the experimental chamber and entered the observation room. The opaque slider was closed after the experimenter had rechecked that the animal stationed correctly and paid attention to the monitor. Then stimulus presentation could be started. In each trial, only one type of stimulus was presented, either the STI or CTI. After the presentation of the respective stimulus for the pre-programmed time, the animal had to indicate its response by touching one of the two response targets. A correct answer was defined as the seal touching the left response target after the STI and the right response target after the CTI. If the response was correct, the slider was opened, and the animal was rewarded with fish. A wrong response was signalled by the experimenter with the German word for no “nein”, the slider stayed closed, and the animal had to station again. The next trial started after approx. 5 s. Only rarely the intertrial interval was prolonged by 2–3 s, for example, if the animal needed to be signalled to pay attention to the monitor or if external noise disturbed the experiment.
One session consisted of 30 trials, 15 of which featured the STI and the other 15 featured the CTI. The sequence with which the two types of stimuli were presented followed a pseudo-randomised scheme (Gellermann 1933).
The time difference threshold of the seals was determined using a modified staircase method. A detailed scheme illustrating the process of threshold determination can be found in Heinrich et al. (2016). In brief, the STI remained constant during the determination of one threshold, and only the CTI was decreased, if the animal had reached the preset learning criterion. The learning criterion was defined as a minimum number of 23 correct trials in a session of 30 trials in total (76.7%, Chi square test: p < 0.01). This performance had to be achieved in two consecutive sessions to meet the learning criterion. If the animal reached the criterion, the duration of the CTI was decreased by either 1 s or 2 s or by halving the time difference between STI and CTI depending on the experimental situation (for details please see Supplementary material). If instead the seal did not reach the learning criterion, up to five sessions were conducted altogether before the threshold determination was ended. Five sessions were conducted as the training had previously revealed that the seal’s performance would not improve even if training was continued for ten sessions. However, if the seal achieved a performance of ≥ 76.7% correct choices in the fifth session, a sixth session was conducted. During this session, the seal could either meet the learning criterion, and threshold determination was continued by decreasing the CTI once more, or its performance again dropped below 76.7 % correct choices, and threshold determination for the respective STI was ended. Only after the determination of a difference threshold for one STI, a new STI was introduced and paired with suitable CTIs for the determination of the difference threshold for the new STI (for details about the sequence with which difference thresholds were determined for specific STIs, please see Supplement).
Analysis
The difference threshold was defined as the time difference between STI and CTI that the animal was able to discriminate with a performance at 75 % correct choices. The difference threshold was calculated via linear interpolation from the mean performance of the last two consecutive sessions above 75 % correct choices and the mean performance of the first five consecutive sessions below 75 % correct choices. Additionally, we calculated the Weber fraction c as
c = \(\frac{\Delta S}{S}\),
with ΔS being the difference threshold for a respective STI, and S, the corresponding STI. The Weber fraction c should be constant for the tested STIs, if Weber’s law is valid for time perception, either for part of or the full range of STIs.
Statistical analysis was performed in IBM SPSS Statistics Version 25 (IBM, Armonk, NY, USA).