Contact: joanna.jia@wecistanche.com / WhatsApp: 008618081934791

Methods

Participants

Seventy-four participants (50 women and 24 men) were recruited at the campus of the Technical University of Darmstadt. Four additional participants did not complete the training sessions (and the post-test) and their data could not be included in the analyses. Ages raged between 18 and 62 years (M = 24.7; SD = 7.5). Participants were ran- domly assigned to either the standard n-back training group (n = 23; 17 women; 19 – 38 years, M = 23.4; SD = 4.3), the inhibitory n-back training group (n = 25; 16 women; 18 – 62 years; M = 25.0; SD = 10.0), or the no-training passive control group (n = 26; 17 women; 18 – 54 years; M = 25.5; SD = 7.0). There were no signifcant age diferences between groups, F(2,71) = 0.48; p = 0.62. All participants reported normal hearing and normal or corrected-to-normal vision. Student participants were compensated with course credits. A sensitivity analysis revealed that the sample size of N = 74 is sufcient to demonstrate an interaction in the present 3 (group) × 2 (pre/post) design with a statistical power of 95% assuming an efect size of f = 0.23 or larger (α = 0.05). Therefore, the present sample size should provide sufcient power to detect the previously reported transfer efects of a dual n-back training on working memory updat- ing (f = 1) (η2p = 0.50 for the interaction using a similar 3 × 2 design; Salminen et al. 2016; p. 10,202) as well as the possi- ble far transfer efect on fuid intelligencef = 0.27 (η2p = 0.07

for the interaction between group and test-session; Jaeggi et al. 2008; p. 6830).

Cistanche can improve memory

Apparatus

The experiment was conducted in a single-walled sound attenuated listening booth (Industrial Acoustics Company, Niederkrüchten, Germany). Visual stimuli were presented on a 17-inch LCD monitor, and participants were seated at approximately 59 cm viewing distance. Sounds were D/A converted at 44.1 kHz (16 bits) by an RME multiface II sound card (Audio Ag, Haimhausen, Germany) and passed through a Behringer HA 800 Powerplay PRO-8 headphone amplifier (Behringer, Zhongshan, China) before being played idiotically via Beyerdynamics DT-990 headphones (Beyerdynamic GmbH, Heilbronn, Germany). The experimental routines were programmed in MATLAB (Mathworks, Natick, MA) utilizing the Psychophysics toolbox 3.0 extensions (Brainard 1997; Kleiner et al. 2007; Pelli 1997).

Procedure

For all three groups, the experiment consisted of a pre-and post-test session. The two training groups (n-back and inhibitory n-back) completed eight 90-min training sessions between the pre-and post-tests, whereas there were no training sessions for the passive control group. The time between pre-and post-tests has intended to be about two weeks, but individual time constraints and preferences resulted in an average interval of 16.6 days (SD = 4.1) in the n-back group, 17.2 days (SD = 6.0) in the inhibitory n-back group, and 10.7 days (SD = 3.8) in the control group. In both training groups, two adjacent training sessions were separated by at least one and no more than two successive days without training.

Pre–post-tests

During the pre-and post-test sessions, participants completed five different tasks in counterbalanced order (using a Latin square design). It took participants about 60 min to read the instructions and complete (a) 25 runs of working memory updating (approx. 10 min), (b) 24 trials of verbal serial recall (approx. 10–15 min), (c) 200 trials of the Simon task (approx. 7 min), (d) 200 trials of task-switching (approx. 11 min), and (e) 18 problems from Raven’s advanced progressive matrices (max. 10 min).

In the working memory updating task (Dahlin et al. 2008a), there were 25 runs in which a variable number of colored circles were presented successively in the center of a black screen (1250 ms per circle, each preceded by a 1000 ms blank screen). For each run, the colors were randomly drawn from nine possible colors (red, green, blue, yellow, pink, cyan, purple, orange, and grey). After a randomly drawn number of trials (between 5 and 12), participants were asked to recall the colors of the last four circles of the series in correct serial order by clicking on the respective colors in a 3 × 3 response matrix showing the nine colors (red, green and blue in the top row, yellow, pink, and cyan in the middle row, and purple, orange, and grey in the bottom row). Clicked responses could not be corrected, and feedback indicating the number of correctly recalled colors was presented on the screen immediately after the fourth response for 1500 ms before the next run started.

In the visual Simon task (Hedge and Marsh 1975), each trial started with a central fixation cross for 500 ms, followed by either a square or a circle (diameter of 3° visual angle) in blue or green color which was presented either 5° to the left or 5° to the right of fixation. Participants were instructed to respond to the color of the stimulus by pressing the ‘A’ key with the left index finger for blue stimuli (left side of the keyboard) and the ‘L’ key with the right index finger for green stimuli (right side of the keyboard), ignoring both the shape and the position of the stimulus. The stimulus-response mappings remained constant across the entire task. In half of the trials, the position of the stimulus was spatially compatible with the response (i.e., a blue stimulus on the left or a green stimulus on the right), and in half of the trials, the position was spatially incompatible (i.e., a blue stimulus on the right or a green stimulus on the left). Participants were instructed to respond as fast as possible while avoid- ing errors. Both the accuracy and the response time (in ms) were shown as feedback after each response that was given within 750 ms (e.g. “Richtig! 326 ms” [Correct! 326 ms]). For response times longer than 750 ms, participants were only prompted to be faster (“Zu Langsam!” [Too slow]). In either case, feedback was presented for 750 ms before the next trial started.

In the verbal serial recall task (Colle and Welsh 1976; Salamé and Baddeley 1982), each trial started with a 1000 ms preparation interval (showing an animated blue square decreasing in size) followed by a random sequence of eight randomly drawn digits (from 1–9). Each digit was presented for 1000 ms, immediately followed by the next digit, and there was an additional 6000 ms retention interval showing a blank screen. To measure the degree of interference by task-irrelevant sound with serial recall, either a passage of free-running Finnish speech (which the participants did not understand) spoken by a male voice (a weather forecast; taken from Kattner and Ellermeier 2014) or white noise was presented during both the presentation and the retention interval. After the retention interval, participants were asked to recall the digits in correct serial order by clicking on the numbers in a 3 × 3 numeric pad shown on the screen. Feed-back indicating the number of correctly recalled digits was presented for 1000 ms before the next trial started (feedback was presented in the green font when 5 or more digits were recalled, and in red when less than 5 digits were recalled). The entire task consisted of 12 trials with irrelevant speech and 12 trials with irrelevant noise which were presented in randomized order.

For the 200 task-switching trials, a random letter (A, E, I, O, K, L, M, or P) and a random number (2–9) were presented on the screen, and participants were asked to categorize either the letter or the number in a typical alternating- runs procedure (Kattner et al. 2019; Rogers and Monsell 1995) providing an equal number of the switch (second and fourth trial) and repeat trials (first and third trial). Specifically, participants indicated whether the letter was a vowel or consonant and whether the number was even or odd by pressing the left or right arrow key on the keyboard. Partici- pants were instructed to respond as fast as possible. Both the accuracy and the actual response time were shown as feedback for 750 ms after each response (e.g. “Falsch! 1689 ms” [Incorrect! 1689 ms]). No feedback but a prompt to be faster was provided if no response was given within 5000 ms.

For both the pre-and the post-test sessions, eighteen unique problems of increasing difficulty were selected from the 36-item short form of Ravens’s Advanced Progressive Matrices (odd item numbers at pre-test and even item numbers at post-test). Each matrix problem was presented on the screen and participants were asked to choose the stimulus which completes the matrix based on a to-be-identified rule. A set of eight response options was shown below the matrix (together with numbers), and responses were made by pressing the respective number on the keyboard. Knowing that the difficulty of problems increases, participants were given ten minutes to complete the eighteen problems one after another without the option to return to a previous problem. The remaining time (in seconds) was shown at the top of the screen. No feedback was provided during or after the task.

Cistanche can improve memory

Training tasks

The n-back group was trained with an adaptive version of the dual n-back task (Jaeggi et al. 2007, 2008; Salminen et al. 2016) for eight 90-min sessions (including breaks and instructions). In this task, a running sequence of visual squares and auditory letters was presented. On each trial, a bluish square (1° diameter) was presented at one of eight possible locations on the screen (at 4° eccentricities from the fixation) while a spoken consonant (B, F, J, K, L, T, or V; male voice) was presented simultaneously via headphones. A new square and a new consonant were presented every second, and participants were asked to hit the space bar on the keyboard whenever one of the present two stimuli (square or letter) matched with the stimuli that were presented n trials back in the sequence (n was the same for squares and letters). No response was required if both the square and the letter were different from the stimuli n trials back. Both the square and the letter were drawn randomly to be identical to the ones n trials back with independent probabilities of p = 0.15 each, so participants should be required to make a response in about 27.7% of the trials, and not to respond on about 72.3% of the trials (based on simulations with the rand function conducted in Matlab). Responses had to be given rapidly within the 1000 ms of stimulus presentation. Text feedback was provided in case of incorrect (‘Finger Weg!’ [Hands of]) or missed responses (‘Verpasst!’ [Missed]). Each training session started with n = 1, but task difficulty was changed with the participants’ performance. Specifically, the value of n was increased by 1 if participants were at least 90% correct in a block of 25 + n trials. The value of n was decreased by 1 if accuracy was at or below 70%, and n did not change if accuracy was between 70 and 90%. Participants could take a short break every 20 min. Each training session continued until either 80 min had passed (break times not included) or 1500 trials were completed (only 13.0% of participants in the n-back training group and no participant in the inhibitory n-back group completed 1500 trials within 80 min).

The inhibitory n-back group was trained on a similar task as the n-back group, with the same random sequences of squares and letters being presented. In contrast to the standard n-back group, participants of the inhibitory n-back group were asked to hit the space key as fast as possible on every trial except when either the location of the square or the spoken letter was identical to the ones n trials back in the sequence. As the independent probabilities of the square and the letter to be identical to the ones n trials back were the same as in the n-back training group (i.e., p = 0.15), participants were supposed to respond in about two-thirds of the trials, and not to respond in about one-third of the trials. Thus, in the inhibitory dual n-back task, pressing the space bar is supposed to be the pre-potent response, which is required on the majority of trials, and participants are required to suppress this response only on the less frequent n-back trials. With equal demands for working memory updating (and other control processes possibly involved), this task is assumed to involve a greater demand for inhibitory control than the standard n-back task.

Results

The individual data from the two n-back pieces of training and the various measures obtained with the tasks at pre-test and post-test are openly available as CSV files in an OSF repository: https://osf.io/ubxap/?view_only=d200d4917bc945c 585913b069631e55a.

Performance during training

Across all participants, 24.6% (SD = 1.4%) of the trials in the n-back group and 24.7% (SD = 1.1%) of the trials in the inhibitory n-back group were “n-back trials” in which either the auditory or the visual item (or both) was identical to the one n trials back. Accordingly, participants in the n-back group responded with a keypress on 23.3% of the trials (SD = 6.3%), whereas participants in the inhibitory n-back group responded with a keypress on 82.1% of the trials (SD = 6.4%). Hence, in contrast to the n-back group, a keypress was the dominant response in the inhibitory n-back group, suggesting that inhibitory control is required to suppress the keypress on the “n-back trials”. Across all eight training sessions, participants of the n-back group completed M = 10,960 n-back trials (SD = 364), whereas participants of inhibitory n-back group completed M = 10,306 trials (SD = 1719), but this difference was not statistically significant, F(1,46) = 3.20; p = 0.08; η2G = 0.06.

Figure 1 illustrates that the average level of n increased across the eight training sessions both in the n-back and

inhibitory n-back groups, demonstrating that the participants’ n-back working-memory updating span increased with both types of training. A 2 (group) × 8 (training session) mixed-factors ANOVA with training session as a repeated-measures factor confirmed this improvement with a significant main effect of training session, F(7,315) = 46.01; p < 0.001; η2G = 0.16. There was no significant main effect of group, F(1,45) = 0.71; p = 0.40; η2G = 0.01, and no group × session interaction, F(7,315) = 0.49; p = 0.84; η2G < 0.01, suggesting that both average performance and the rate of learning did not differ between the two training

groups.

cistanche bodybuilding

Transfer to working memory updating

Transfer effects of the two types of training were first assessed with regard to the working memory span in an untrained color updating task. The average number of correctly recalled colors in the updating task is illustrated in Fig. 2 for the three experimental groups. Intra-class correlations between the average working memory updating span at pre-and post-test (calculated with the {irr} package in R) demonstrated good test-retest reliability, ICC(C,2) = 0.73 (N = 74). Interestingly, all groups significantly improved their color updating span from pre-test (M = 2.64; SD = 0.51) to post-test (M = 3.09; SD = 0.46), F(1,71) = 79.35; p < 0.001; η2G = 0.18. However, the improvement also differed between groups, as suggested by the significant interaction, F(2,71) = 3.25; p = 0.04; η2G = 0.02. Pairwise t-tests on the increments of working memory updating spans, corrected for multiple comparisons (Benjamini and Hoch- berg 1995), revealed a significant difference between the inhibitory n-back group and control group, p = 0.04, but no difference between the inhibitory n-back and standard n-back groups, p = 0.17, and the standard n-back and control groups, p = 0.41. In addition, Bayes factors (calculated with the R package {BayesFactor}; Rouder et al. 2009) indicate that it is about three times more likely that the inhibitory n-back produced a greater increment of updating span than in the control group (BF10 = 3.08), whereas the efect of the standard n-back training does not seem to differ from the control group (BF10 = 0.37). This suggests that the enhanced demands for inhibitory control during training in the inhibitory n-back group produced more transfer to general updating abilities than the training with the standard dual n-back task. There was no main efect of group on the updating span, F(2,71) = 0.41; p = 0.66.

cistanche and tongkat ali