Participants
One hundred and seventy-one student participants (age: mean (M) = 21.719, standard deviation (SD) = 2.167; gender: 43 males and 128 females) were recruited from the National Cheng-Chi University, Taipei, Taiwan. All of them were self-reported native Chinese speakers, right-handed, with normal or corrected-to- normal vision, not a heavy drinker or taking any medication, and without any history of neurological or psychiatric disorders. Written informed consent was obtained from each participant before the experiment, and NT$150 was reimbursed after the experiments. All methods were performed in accordance with the ethics principles of Declaration of Helsinki, and the study was approved by the Research Ethics Committee of National Cheng-Chi University.
Experimental procedures and materials
Upon arrival at the laboratory, each participant was introduced to the aim of this study in a sound-attenuated room and signed an informed consent form if they agreed to participate. They were told to complete three behavioural tasks conducted using a computer and a handwritten self-report questionnaire. The designs of these measurements, the DRL, SSRT, TPT, and the Chinese version of BBIS-11 (Li & Chen, 2007) are described below.
Differential reinforcement of low-rate responding task-10 seconds (DRL-10s)
In the DRL-10s, participants were instructed that their goal was to try and earn as many green circles as possible by pressing a space button, but were not given how to do so exactly from the instruction. This design provided a better ecological analogue to previous animal DRL studies and excluded the influence of humans’ prior knowledge or experience 63. The more the green circles were shown on the computer screen, the higher the opportunities for participants to receive accumulated rewards. In the task, there were six sessions, each lasting 5 min. At the beginning of each session, a white fixation cross “+” was shown at the centre of the screen without a time limit, and the participant could click the left mouse button when he/she was ready to start the experiment. After clicking, a yellow square was displayed at the centre of the screen as the signal to start and maintained for 5 min, during which each participant could try multiple ways of pressing the space button on a keyboard. A green circle would be shown at the centre of the screen for 1 s as a reinforcement if participants pressed the button within the time interval of 10–13.999 seconds (10s with a limited hold of 3.999s) since the previous response occurred. Conversely, none of the reinforcements would be shown if their response was beside the reinforced time interval (IRT < 10s and >14s), and the yellow square would continue to be displayed until the participants learned how to earn the reinforcement by pressing the space button. Before performing the task, each participant underwent a continuous reinforcement schedule [fixed-ratio (FR) 1] with six trials as practices to learn the association between pressing a space button to an orange square and receiving a green circle on the screen as reinforcement. The total duration of performing DRL-10s was 30–40 min (see Figure 1(A)).
Stop-signal reaction task (SSRT)
In the SSRT, participants were instructed that their goal was to press the space button when the go signal appeared but withhold their response when the stop signal appeared as fast and accurately as possible. There were three blocks, each consisting of 50 trials. At the beginning of each trial, a white fixation cross “+” was shown at the centre of the screen for 1–2 s, followed by a short black screen for 200 ms. Then, a square sign was displayed at the centre of the screen for 500 ms as the go signal, to which participants had to press the space button as soon as possible. However, after displaying the go signal, there was a 30% probability to appear the two squares side by side for only 75 ms as the stop signal, and the duration between displaying the go and stop signals was only 200, 250, or 300 ms randomly. Under this condition, participants had to withhold their already-initialised pressing responses as quickly as possible. The duration of each trial was 1250 ms regardless of whether each executing response was correct, and the inter-trial interval (ITI) varied from 1500 to 3500 ms. Before performing the task, there were six trials, including four go-signals and two stop-signal trials, as practice for participants to familiarise themselves with the task. The total duration of the SSRT was 10–15 min (see Figure 1(B)).
Time production task-10 seconds (TPT-10s)
In the TPT-10s, participants were instructed that their goal was to produce a particular duration (10 s) by pressing a space button in each trial as precisely as possible. There was only one block with 30 trials for this task. At the beginning of each trial, a white fixation cross “+” was shown at the centre of the screen for 1–2 s; then, it was replaced by a blue square sign as the start signal. During each trial, the participants had to count to 10 s themselves and press the space button as a response. Compared to the design of DRL-10s, participants instructed with TPT-10s would clearly discern that the correct response was defined by producing a pressing response within 10 to 10.999 s since the square sign had been displayed. After responding, the precise response time would be shown at the centre of the screen for 2 s regardless of whether it was correct (e.g., 10766 or 8943 ms). This allowed each participant to adjust his/her timing production based on the provided feedback. The inter-trial interval (ITI) was designed as a black screen of 1400, 1500, or 1600 ms presented randomly. Before performing, six practice trials required participants to count 7 s (TPT-7s) and press the space button to familiarise them with the task procedure. The total duration of the TPT-10s was 8–10 min (see Figure 1(C)).
Chinese version of Barrett impulsive scale, version 11 (BIS-11)
BIS-11 is a 30-item self-report questionnaire designed to assess the attentional (the inability to focus or concentrate), motor (the tendency to act without thinking), and non-planning (the tendency to plan without futuring or foresight) impulsivity 64,65, which has already been translated to various languages with good internal consistency and test-retest reliability e.g. 66,67-69. The Chinese version of the BIS-11, translated by Li and Chen 43, used a four-point scale (1 = rarely/never; 2 = occasionally; 3 = often; 4 = almost always/always) to measure, so the total score ranged from 30 to 120. Participants who score higher are considered with higher impulsivity 43,64,65,70. The internal consistency coefficient (Cronbach’s α) for all items was .78 in the study by Wang and Yu 71. The total duration of answering the Chinese version of the BIS-11 was 3–5 min.
(insert Figure 1 about here)
The order of the three tasks was randomly presented to each participant, with the DRL-10s always being administered prior to the TPT-10s, to prevent participants from inferring that the reinforced responses in the DRL might be related to timing behaviour. After completing all tasks, participants answered the BIS-11 and received a participation fee of NT$150 after the experimenter provided debriefing. Three of the participants did not complete the entire experiment due to their personal arrangements, and one participant’s data was missing due to a recording error on the computer. Data from the remaining 167 participants were analysed. The entire experiment lasted 60–70 min.
Data collection
The DRL behaviour data were based on the pressing responses recorded by the inter-response time (IRT; the time elapsed since the previous response), and six indices were calculated: (1) the number of total responses, (2) the number of reinforced responses, (3) the number of non-reinforced responses, (4) the number of burst responses, (5) the peak rate, and (6) the peak time. The first three indices were frequency measures, which accumulated the number of total pressing responses, the number of pressing responses within IRT10-13.9999s, and the number of pressing responses within both IRT < 10s and > 14s during DRL. The burst response was the sum of the number of pressing responses within IRT < 2s, reflecting response disinhibition. The peak rate and peak time were calculated from IRT > 2s, where a moving average based on four consecutive 1-s bins with 1-second step size was applied to smooth the distribution. The peak rate was calculated as the summed number of pressing responses in the four bins divided by four, which rendered a unit of responses per second for the peak rate, reflecting an individual’s motivation to perform the DRL task. After the maximum of the summed pressing responses of a four-second epoch was identified, the peak time was the average value in ms of all IRTs that fell within those four bins (i.e. the maximal epoch), which showed the time point where participants pressed the button with the highest numbers, reflecting their expected criterion time in DRL see 29,72. In SSRT, the false-responding rate from only stop-signal trials, namely the percentage of false-pressing reactions, was calculated as the index of response disinhibition. In TPT, the response accuracy, that is, the percentage of pressing responses within the time range of 10–10.999s was calculated as the index of time estimation.
Statistical analysis
To investigate how participants performed the DRL, individual differences were first examined. Participants were divided into high efficiency ratio (HE) and low efficiency ratio (LE) groups according to their efficiency performance; that is, the number of reinforced responses divided by that of the total responses. A two-way mixed ANOVA, with the group (HE and LE) as a between-subject factor and session (six sessions) as a within-subject factor, was conducted on the behavioural indices of DRL. Second, we tested the hypothesis that the underlying behaviour of DRL was response inhibition or timing estimation. Pearson’s correlation test was conducted between the four indices of DRL (efficiency ratio, burst responses, peak rate, and peak time), the false rate of SSRT, the accuracy of TPT, and the subjective scores of BIS-11 (including the total impulsive score and the three subscale scores). In Addition, principal component analysis (PCA) using varimax rotation was conducted on the same measurements. The outcomes of the yielding components with eigenvalues ≥ 1 and loadings ≥ 0.3 were considered critical for interpretation. It should be noted that the distribution of the original peak time was bimodal; hence this measure was transformed by coding the value ≧ 8.2584 (the mean of the original peak time distribution) as 1 and < 8.2584 as 0, making the data as continuous variables for conducting the correlational test and PCA. All analyses were performed using SPSS version 21.