Maximum load or the decay rate of memory 

Now that I’ve come to understand how Interference Theory (IT) and Cognitive Load Theory (CLT) function in supporting meditative absorption (samādhi), the next question is: what do I do with that understanding? Should I try to identify a universal threshold—like a standard amount of time or a specific number of mental steps—that leads to samādhi? 

From months of personal experience, I’ve found that simply generating three to six new numbers [1]from a set of initial figures is enough to hit the threshold of WM capacity. This range varies depending on many factors. But if we just focus on the characteristics of the initial information, the key determinants include: the type of characters used (letters like a, f, z, etc.; Roman numerals like 2, 5; Latin numerals like I, IV, LX, etc.); the font style (curved, straight, jagged, etc.); the length of the numbers (e.g., 9 vs. IX); the color complexity (continuous, disjointed, overlapping); and the semantic content (how meaningful, readable, or rhythmically memorable the elements are). All these factors determine how much of WM’s capacity is consumed at the start. Then, the remaining load is taken up by the effort to recall, compute, arrange, and generate new numbers, which together determine how many steps I can create before WM collapses into stillness. This is not to mention indirect factors like posture, bodily sensations, mental states, or karmic forces. Therefore, each person must practice, observe, and calculate for themselves what their own WM limit range is. 

Even so, it’s worth briefly reviewing some scientific studies for reference. [2] WM load can generally be measured in two ways: max load (i.e., how much information causes overload), or decay rate (i.e., how quickly memory fades from 100% to 0%). These two perspectives are like the age-old question of whether a glass is half full or half empty. Researchers have devised many formulas to measure WM load, but in my view, most haven’t succeeded—certainly not with meditation in mind .

Among the limited materials I’ve read, here are some notable findings: 

+ Regarding the maximum number of information units (chunks) WM can accurately retain Miller (1956)[3] proposed the classic figure of 7 ± 2, i.e., ranging from 5 to 9. Cowan (2001)[4] through a more rigorous synthesis of post-Miller studies, argued the actual range is closer to 4 ± 1, or 3 to 5 chunks. 

+ Hancock và Chignell (1988)[5] proposed a way to estimate cognitive load using three variables: e effort required, t effective time allocated to the task, and s personal skill level. Their model wasn’t grounded in a theoretical framework. 

+ Bi và Salvendy (1994a, 1994b)[6] borrowed analogies from physical material load to estimate mental workload, using four known parameters: l (lamda) task occurrence rate (e.g., calculation steps), T_ui unpredictability of the next task, T_cij task complexity, and P_i system busyness (e.g., time pressure or quality demands). They also included environmental factors (K_e living environment and K_v working environment). However, they didn’t explain how these variables might interact with one another.[7] 

+ Patel, Salvendy, Geddes, và Kuczek (2002)[8] used Ohm’s Law from electrical engineering to model cognitive load: current (I) represented the information transmission rate, voltage ( V) equated to the mental load per unit of information, resistance (R) ) included task-related obstacles (complexity, time constraints) and personal hindrances (fatigue, cognitive capacity, anxiety, pressure, etc.) Their empirical results showed that as I increases V decreases—and the higher the resistance (R), the less effectively one processes information, since V is lowered. But the model was overly abstract, lacking clarity and failed to show clear interactions between I and R. 

+ Rubin, Hinton, và Wenzel (1999)[9] synthesized various models for estimating memory retention (or, inversely, forgetting). These models included: logarithmic functions, power/exponential functions, and square root functions. They proposed a model integrating both short-term and long-term memory. Later, Murre and Dros (2015)[10] introduced an improved version called the Memory Chain Model. Focusing on short-term memory (WM), one can see that retention probability Q(t) depends on three variables: m memory trace strength, a decay rate, and t elapsed time (Note: this model doesn’t account for intervening items.) Given the equation Q(t) = me^-at, assuming a = 1, when t = 1 , Q(1) = me^-1*1 = 0.37m, only 37% of the original memory trace remains. When t = 2 thì Q(2) = me^-1*2 = 0.14m, just 14% retention left. 

In summary, after more than a century of research, scholars still haven’t found a complete model to accurately measure cognitive load. Most of the models I’ve encountered seem like reverse-engineered attempts to fit existing data. Human beings are simply too diverse. And in the context of meditation, it’s just “me with myself.” So I believe the best approach is to practice, experiment, observe, and derive your own personal estimate of WM’s capacity.. 

(End of Part 5/11)

Notes:

[1] Brother Alpha, during his own practice, found that 8 was the number that led to samādhi. However, I don’t recall whether this referred to eight starting numbers, or eight newly generated numbers.

[2] Research results vary—sometimes aligning, sometimes diverging—depending on factors such as the participant profile, measurement tools, and methodology used. Moreover, the results are usually reported as average values, but the variation around the average (standard error), the model’s reliability (e.g., R-squared, F-statistic), and other statistical indicators are also important considerations. 

[3] Miller, G.A., 1956. The magical number seven, plus or minus two: Some limits on our capacity for processing information. The Psychology Review 63(2), 81–97. 

[4] Cowan, N., 2001. The magical number 4 in short-term memory: A reconsideration of mental storage capacity. Behavioral and Brain Sciences 24, 87–185. 

[5] Hancock, P.A., and Chignell, M.H., 1988. Mental workload dynamics in adaptive interface design. IEEE Transactions on Systems, Man and Cybernetics 18(4), 647–658. Trang 651: W = 1/et^s-1. 

[6] Bi, S., and Salvendy, G., 1994a. Analytical modeling and experimental study of human workload in scheduling of advanced manufacturing systems. The International Journal of Human Factors in Manufacturing 4(2), 205–234. Bi, S., and Salvendy, G., 1994b. A proposed methodology for the prediction of mental workload, based on engineering system parameters. Work & Stress 8(4), 355–371. 

[7] Although their model explained 60% of the experimental data, the authors acknowledged that much more real-world data would be needed to assess the model’s reliability. See page 215 in BS (1994a) and page 367 in BS (1994b).

[8] Patel, U.H., Salvendy, G., Geddes, L.A., Kuczek, T., 2002. An electrical-circuit model for predicting mental workload in computer-based tasks. Journal of the Chinese Institute of Industrial Engineers 19(1), 1–15. Trang 5, V = IR. 

[9] Rubin, D.C., Hinton, S., and Wenzel, A., 1999. The precise time course of retention. Journal of Experimental Psychology: Learning, Memory, and Cognition 25(5), 1161–1176. [10] Murre, J.M.J., and Dros, J. 2015. Replication and analysis of Ebbinghaus’ forgetting curve. PLOS ONE 10(7), 1–23. See full description on page 18.