Short-term memory of temporal information revisited

The final sample consisted of 48 participants (36 women) with a mean age of 23.5 (SD = 3.6) years. Each participant took part in two separate sessions of the experiment (each lasting about 45 min). They received either course credit or payment for their participation. Eight original participants were replaced by new participants because they did not adhere to the instructions or achieved an accuracy in the interference tasks at or below chance level (≤ 53% correct).

The experiment was programmed in MATLAB^® using the Psychophysics Toolbox extension (Brainard, 1997; Kleiner et al., 2007). Participants sat in front of a computer screen with a viewing distance of approximately 50 cm. Auditory stimuli were presented binaurally via headphones (Sennheiser HD 380 pro) with a peak amplitude of 60 dB (A) SPL. White noise (400, 440, 480, 520, 560, and 600 ms) with rise and fall times of 5 ms served as the auditory duration stimulus in the duration discrimination task. In the pitch discrimination task, there was a reference sine tone of 800 Hz and 24 test sine tones ranging from 320 to 1280 Hz (in steps of 40 Hz with the exception of 800 Hz).

Visual stimuli were presented on a 21-inch CRT monitor (Samsung SyncMaster 1100 MB, 1024 × 768 pixels, 150 Hz). To make the visual duration stimuli as similar as possible to the auditory ones, we used a square (11.4 × 11.4° of visual angle) filled with dynamic grey visual noise (random grey pattern of 256 × 256 pixels with 150 Hz refresh rate) as the visual duration stimulus. As in Bratzke et al. (2016), we used a wider range of durations for the visual (320, 380, 440, 560, 620, and 680 ms) than for the auditory stimuli (400–600 ms) to compensate for the typically observed poorer temporal sensitivity for visual than auditory stimuli (e.g., Bratzke & Ulrich, 2019b; Grondin, 1993; Ulrich et al., 2006). A violet-colored (RGB values: 122, 24, 122) square of the same size (11.4 × 11.4° of visual angle) served as reference stimulus in the color discrimination task. There were 24 test colors, of which 12 can be categorized as more reddish or bluish than the reference color. The RGB values of the reddish stimuli ranged from (190, 11, 55) to (135, 22, 110) and those of the bluish stimuli ranged from (110, 22, 135) to (55, 11, 190), with steps of five units for the R and B values and 1 unit for the G value. All visual stimuli were presented at the center of the screen on a white background. Responses were collected using the ‘X’ (left index finger) and ‘M’ (right index finger) keys on a standard German keyboard.

Each participant took part in two experimental sessions, a visual and an auditory discrimination session (with a maximum separation of one week between the two sessions). Each trial started with the presentation of a small fixation point at the center of the screen. After 750 ms, the standard duration was presented, chosen randomly with equal probability from the six possible durations. Then, an 8-s retention interval started with a blank screen. 750 ms before the end of the retention interval the fixation point reappeared and remained at the center of the screen until the presentation of the comparison duration. After the retention interval, the comparison duration was presented. This stimulus was either equal to the standard duration (50% “same” trials) or it differed (50% “different” trials). In “different” trials, when the standard duration was shorter than 500 ms, the comparison duration was 120 (240) ms longer in auditory (visual) trials. When the standard duration was longer than 500 ms, the comparison duration was 120 (240) ms shorter in auditory (visual) trials. 1000 ms after the end of the comparison duration, a prompt message was presented asking the participant to indicate whether the durations of the two stimuli were equal or different. The message included a reminder of the response-to-key key assignment in the duration discrimination task. The next trial started 1000 ms after the participant’s response. For the duration discrimination task, participants were instructed to respond as accurately as possible without any speed restrictions.

Within each modality session (visual vs. auditory duration discrimination), participants completed three blocks of trials, each differing in the nature of the interference task presented during the retention interval. In “no interference” trials, the screen remained empty during the first 7.25 s of the retention interval. In color discrimination trials (visual interference), the reference-colored square appeared after a random interval between 1 and 2 s from the start of the retention interval. After another random interval between 1 and 2 s, the square changed its color to one of the 24 test colors. Participants had to indicate whether the color changed toward more reddish or bluish. Irrespective of the participant’s response time, the test color stimulus remained on the screen for another random interval between 1 and 2 s. Pitch discrimination trials (auditory interference) followed the same time course as visual ones. In these trials, first the reference tone was played, which then changed to one of the 24 possible test tones. Participants had to indicate whether the tone’s pitch changed toward higher or lower. For the interference tasks, participants were instructed to respond as quickly and as accurately as possible to the change of the reference stimulus.

For each of the combinations of duration modality (visual vs. auditory) and interference task (no interference, color discrimination, and pitch discrimination), participants completed three practice trials (randomly chosen from all possible trials within each factorial combination) and four experimental blocks of 12 trials each (one “same” and one “different” trial for each of the possible 6 durations). Within each experimental block, all durations were presented equally often and the order of trials was randomized. After each experimental block, participants received feedback about the percentage of correct responses in the duration discrimination task in the preceding block. Duration modality varied between sessions, and interference task varied between blocks within each session. The order of modality sessions and interference tasks was balanced across participants. The response-to-key assignment in the duration discrimination task was also balanced across participants. The response-to-key assignment in the interference tasks was the same for all participants (color discrimination: more bluish—‘X’/more reddish—‘M’; pitch discrimination: lower—‘X’/higher—‘M’).

Figure 2 depicts \(A^{\prime}\) (upper panel) and \(B^{\prime\prime}\) (lower panel) values as a function of duration modality and interference task. Note that \(A^{\prime}\) can range from 0 to 1, with \(A^{\prime}\) = 0.5 indicating chance level, and \(A^{\prime}\) = 1 representing perfect discrimination performance. ANOVA revealed a significant main effect of interference task, F(2, 94) = 5.56, p = 0.005, \({\eta }_{p }^{2}= 0.11\). Post hoc Tukey contrasts indicated a reliable difference between pitch discrimination and no interference (0.71 vs. 0.76), p = 0.006, a marginally significant difference between pitch and color discrimination (0.71 vs. 0.74), p = 0.077, and no significant difference between color discrimination and no interference (0.74 vs. 0.76), p = 0.639. This pattern suggests that only pitch discrimination interfered with short-term memory of temporal information. The main effect of duration modality was also significant, F(1, 47) = 11.06, p = 0.002, \({\eta }_{p }^{2}= 0.19\). Discrimination performance was on average slightly better for visual (0.76) than for auditory (0.72) duration stimuli. A paired sample t-test on discrimination performance only in the control condition revealed no difference between visual and auditory stimuli (0.76 vs. 0.75), t(47) = 0.63, p = 0.531. Most important, the interaction between duration modality and interference task was also significant, F(2, 94) = 6.43, p = 0.002, \({\eta }_{p }^{2}= 0.12\). Separate ANOVAs for the two duration modalities revealed that the interference tasks impaired discrimination performance when the duration stimuli were auditory, F(2, 47) = 11.02, p < 0.001, \({\eta }_{p }^{2} = 0.12\), but not when the stimuli were visual, F(2, 47) = 0.07, p = 0.934, \({\eta }_{p }^{2}< 0.01\). Post-hoc Tukey contrasts indicated that pitch discrimination differed from color discrimination (p = 0.002) as well as no interference (p < 0.001), and that color discrimination did not differ from no interference (p = 0.567). Together, these results show a selective interference pattern: only the auditory interference task (pitch discrimination) impaired discrimination performance and this was only the case for auditory duration stimuli.

In contrast to the previous results by Bratzke et al. (2016) and Rattat and Picard (2012), most of the \(B^{\prime\prime}\) values were slightly positive (see lower panel of Fig. 2), indicating a small bias towards “different” responses (\(B^{\prime\prime}\) values can range from − 1 to + 1, with negative values indicating a bias towards “same” responses and negative values indicating a bias towards “different” responses). There was a significant main effect of duration modality, F(1, 47) = 20.97, p < 0.001, \({\eta }_{p }^{2}=0.31\), with a bias for visual (0.12) and virtually no bias for auditory (0.01) duration stimuli. The main effect of interference task (p = 0.091) and the interaction between duration modality and interference task (p = 0.092) were not significant.

Figure 3 depicts mean RT and error rates as a function of interference task and duration modality. The analysis of mean RT revealed a main effect of duration modality, F(1, 47) = 10.69, p = 0.002, \({\eta }_{p }^{2}= 0.19\). Responses to the color or pitch change were faster for visual (651 ms) than for auditory (765 ms) duration stimuli. The main effect of interference task was not significant, F(1, 47) = 2.21, p = 0.144, \({\eta }_{p }^{2}= 0.01\). There was also no indication of any interaction between duration modality and interference task, F(1, 47) = 0.03, p = 0.854, \({\eta }_{p }^{2}< 0.01\). The analysis of error rates revealed no significant main or interaction effect, all ps ≥ 0.123. Secondary task performance thus shows a pattern slightly different to the one of primary task performance. Auditory duration discrimination impaired response speed in the secondary task, irrespective of the modality of the interference task¹.