5D Library › Neurobiology of Ritual, Spatial Boundary Transitions and Computational Biomedicine of Self-Transcendence · 中文

Chapter 1 More Than Superstition, Less Than a Cure-All: Anxiety, Rigid Action and Precision Weighting

What ritualised behaviour is

The author’s framework opens with a claim: ritual is “not mere superstition”. To test it we first need to say what “ritual” means. In their long article in Behavioral and Brain Sciences, Boyer and Liénard (2006) listed the features of ritualised behaviour: the action sequence is rigid and must be performed in a fixed order and manner; it contains much redundancy and repetition; there is no direct causal link between the actions and their stated purpose (“goal demotion”); and leaving it undone causes unease (Boyer and Liénard 2006). They argued that such behaviour engages a “precaution system”, alert to potential, vague threats (contamination, disorder, exclusion), and that a rigid sequence fills attention so that thoughts of threat temporarily leave working memory. The 2018 review by Hobson and colleagues divided the psychological functions of ritual into regulating emotion, regulating performance goals and regulating social connection, and stressed that most of these functions rest on small laboratory studies (Hobson et al. 2018).

Two questions with different directions must be kept apart. The first is whether anxiety elicits ritualised behaviour; the second is whether ritual reduces anxiety. The framework cites Lang and colleagues (2015) in support of the second question, but that study answers the first.

Evidence

Anxiety elicits ritualised behaviour (human experiment). In an experiment in the Czech Republic, Lang, Krátký, Shaver, Jerotijević and Xygalatas randomly assigned 62 university students to two groups. The high-anxiety group prepared a speech about a decorative metal object and delivered it to a panel of experts; the low-anxiety group prepared a speech but did not have to give it. Both groups then cleaned the object while an accelerometer on the wrist recorded their hand movements and their heart rate was monitored. The high-anxiety group cleaned for longer and repeated the same hand movements more often; their action sequences were more rigid and redundant (Lang et al. 2015). The authors present this as experimental evidence that anxiety elicits ritualised behaviour and suggest that ritualisation may be a coping strategy; but the study did not test whether anxiety fell after ritualising.

Evidence

Ritual reduces anxiety (human experiments with modest samples). Brooks, Schroeder, Risen, Gino, Galinsky, Norton and Schweitzer (2016) reported several experiments in which completing a short ritual designed by the researchers before singing in public or doing arithmetic was followed by lower self-reported anxiety and heart rate and by better performance; the same actions helped more when described as a “ritual” than as “random behaviours” (Brooks et al. 2016). Norton and Gino (2014) reported that people who recalled or performed mourning rituals after losing a loved one, a lover or a lottery felt less grief and more control (Norton and Gino 2014). These studies show that, in the laboratory, ritual can bring moderate improvements; but their samples were usually a few dozen per group, the “rituals” were devised by the researchers, and instructions and expectations mattered. In checking for this book we found no large pre-registered replications of these findings.

Evidence

A pre-registered test (human experiment). Karl and Fischer (2018) ran a pre-registered study (registered on the Open Science Framework) analysing 180 undergraduates (Karl and Fischer 2018). In the direction “stress makes movement more repetitive”, the results partly replicated Lang and colleagues: stressed participants cleaned with more repetitive movements. In the direction “ritual reduces stress”, the results were essentially null: neither the cleaning movements alone nor a cognitive load alone (reciting) lowered self-reported anxiety or physiological arousal; only the two combined were followed by a slightly larger fall in heart rate during the intervention, which the authors cautioned against over-interpreting; the links between repetitiveness and falls in heart rate and diastolic pressure were only marginal (p=.09p = .09 and .08.08). The planned test of a “novel ritual” was not carried out because the preceding hypothesis failed. The study shows that the evidence for “ritual reduces anxiety” is much weaker than for “anxiety elicits ritual”.

So the claim “ritual is mere superstition” can be falsified: in controlled experiments there is a reproducible relation between ritualised behaviour and anxiety, and ritual is sometimes accompanied by lower anxiety and better performance. But falsifying one claim does not establish its opposite. Over-correction must be avoided: the effects are modest, depend on instructions and expectations, and published effects from small studies tend to be inflated.

Small samples and the winner’s curse

To illustrate the last point, the next program runs a purely simulated exercise: suppose a ritual really does lower anxiety, with a true standardised effect dd of 0.2, 0.35 or 0.5; many laboratories each run a two-group comparison with 20, 40 or 100 people per group; only significant results are published.

Calculation

# Small true effects, small samples and the winner's curse (Chapter 1). Synthetic only.
# Suppose a ritual lowers self-rated anxiety by a true standardised effect d. Many labs
# run two-group studies with n per group; only "significant" results (two-sided p < 0.05,
# normal approximation) are published. What is the power, and how large is the average
# published effect? How many participants would 80 % or 90 % power need?
import math, random, statistics

ND = statistics.NormalDist()
Z = ND.inv_cdf

def power(d, n, alpha=0.05):
    se = math.sqrt(2 / n); z = Z(1 - alpha / 2)
    return 1 - ND.cdf(z - d / se) + ND.cdf(-z - d / se)

def n_needed(d, pw, alpha=0.05):
    return math.ceil(2 * (Z(1 - alpha / 2) + Z(pw)) ** 2 / d ** 2)

def mv(x):
    m = math.fsum(x) / len(x)
    return m, math.fsum((v - m) ** 2 for v in x) / (len(x) - 1)

def lab_study(rng, d, n):
    (ma, va), (mb, vb) = mv([rng.gauss(d, 1) for _ in range(n)]), mv(
        [rng.gauss(0, 1) for _ in range(n)])
    est = (ma - mb) / math.sqrt((va + vb) / 2)
    return est, abs(est) / math.sqrt(2 / n) > Z(0.975)

rng = random.Random(2016)
print("true d   n/group   power   share published   mean published d   exaggeration")
for d in (0.2, 0.35, 0.5):
    for n in (20, 40, 100):
        res = [lab_study(rng, d, n) for _ in range(2000)]
        pub = [e for e, s in res if s and e > 0]
        mp = statistics.fmean(pub)
        print(f"  {d:4.2f} {n:8d} {power(d, n):8.2f} {len(pub) / len(res):14.2f}"
              f" {mp:17.2f} {mp / d:13.2f}x")
print("participants per group for a two-sided test at alpha 0.05:")
for d in (0.2, 0.35, 0.5):
    print(f"  d = {d:.2f}: 80 % power {n_needed(d, 0.8):5d};"
          f" 90 % power {n_needed(d, 0.9):5d}")
# true d   n/group   power   share published   mean published d   exaggeration
#   0.20       20     0.10           0.11              0.79          3.96x
#   0.20       40     0.15           0.15              0.56          2.78x
#   0.20      100     0.29           0.30              0.37          1.86x
#   0.35       20     0.20           0.20              0.80          2.29x
#   0.35       40     0.35           0.35              0.60          1.70x
#   0.35      100     0.70           0.71              0.42          1.21x
#   0.50       20     0.35           0.35              0.85          1.69x
#   0.50       40     0.61           0.62              0.65          1.30x
#   0.50      100     0.94           0.93              0.52          1.04x
# participants per group for a two-sided test at alpha 0.05:
#   d = 0.20: 80 % power   393; 90 % power   526
#   d = 0.35: 80 % power   129; 90 % power   172
#   d = 0.50: 80 % power    63; 90 % power    85

With a true effect d=0.2d = 0.2 and 20 per group, power is only 0.10, and the published studies report on average d=0.79d = 0.79, 3.96 times the true effect; with 100 per group the inflation is still 1.86-fold. Only when power is high (d=0.5d = 0.5, 100 per group, power 0.94) does the average published effect come close to the truth (1.04-fold). Detecting d=0.35d = 0.35 with 90 % power needs 172 per group; detecting d=0.2d = 0.2 needs 526. This does not say that any particular ritual study overestimated its effect; it says that in a field where samples are a few dozen and publication bias exists, the effect sizes of the first studies should be discounted until large pre-registered replications arrive. Book 33 discussed the statistics of this phenomenon; one example is enough here.

Predictive coding and precision weighting: a brief recap

The second step of the framework explains ritual with Friston’s active inference and predictive coding. Book 15 discussed variational free energy, precision weighting and Markov blankets in detail; only the point needed here is recalled (Friston 2010). In predictive coding, each level of the brain explains the input from below with predictions from above and passes only prediction errors upward; errors are weighted by their precision (inverse variance). For a Gaussian prior N(μp,1/πp)N(\mu_p, 1/\pi_p) and a Gaussian observation xx with precision πs\pi_s, the posterior mean is

μpost=wx+(1−w)μp,w=πsπs+πp.\mu_{\text{post}} = w\,x + (1-w)\,\mu_p, \qquad w = \frac{\pi_s}{\pi_s + \pi_p}.

The higher the sensory precision, the closer the posterior is to the observation; the higher the prior precision, the closer it is to the prior. In this framework anxiety is often described as overestimated “environmental uncertainty”, or as excessive trust in a prior of threat.

The author’s model: rigid sequences and a cascade of bias

The framework says that rigid, low-entropy (H→0H \to 0) action sequences raise proprioceptive precision and so damp the “cascading amplification” of top-down prior errors. Below, this claim is written as a computable model. It is the author’s model, and each step is an assumption.

Step one measures the uncertainty of the action sequence. There are 6 actions; with probability rr the next action follows a fixed “script”, otherwise it is drawn uniformly from the six. The larger rr, the more rigid the sequence. We measure the conditional entropy HH (bits) of the next action given the current one and the Gini impurity G=1−∑pi2G = 1 – \sum p_i^2 (the expected squared error of the best one-hot forecast).

Step two assumes sensory precision πs=1/(s0+G)\pi_s = 1/(s_0 + G): a predictable sequence produces small prediction errors and hence high precision.

Step three assumes that a top-down bias of size 1 (say the prior “I am going to embarrass myself”) passes down LL levels; at each level the posterior keeps a share (1−w)(1-w) of the bias from above, and the bias is multiplied by a gain gg (the “cascade”). The bias reaching the bottom is (g(1−w))L\bigl(g(1-w)\bigr)^L. It is damped when g(1−w)<1g(1-w) < 1, that is w>1−1/gw > 1 – 1/g.

Calculation

# The author's model of a rigid action sequence and precision weighting (Chapter 1).
# A: six actions; with probability r the next action follows a fixed script, otherwise it
#    is drawn uniformly. We measure the conditional entropy H (bits) of the next action
#    given the current one, and the expected squared error of the best one-hot forecast
#    (Gini impurity G). Sensory precision is modelled as pi_s = 1 / (s0 + G).
# B: a top-down bias of size 1 passes down L levels. At each level the posterior keeps a
#    share (1 - w) of the bias from above, w = pi_s / (pi_s + pi_p), and the bias is
#    multiplied by a gain g (the "cascade"). The bias reaching the bottom is (g(1-w))^L.
# C: the "dark room" caveat: the same model lowers its estimate of environmental
#    volatility when it is fed its own predictable actions, though the world is unchanged.
import math, random

ACTIONS, SCRIPT = 6, [1, 2, 3, 4, 5, 0]          # scripted successor of each action

def sequence(rng, r, n):
    a, out = 0, [0]
    for _ in range(n - 1):
        a = SCRIPT[a] if rng.random() < r else rng.randrange(ACTIONS)
        out.append(a)
    return out

def cond_entropy(seq):
    pair = [[0] * ACTIONS for _ in range(ACTIONS)]
    for x, y in zip(seq, seq[1:]):
        pair[x][y] += 1
    n, h, g = len(seq) - 1, [], []
    for row in pair:
        tot = math.fsum(row)
        if tot == 0:
            continue
        p = [c / tot for c in row]
        h.append(tot / n * -math.fsum(q * math.log2(q) for q in p if q > 0))
        g.append(tot / n * (1 - math.fsum(q * q for q in p)))
    return math.fsum(h), math.fsum(g)

S0, PI_P, GAIN, LEVELS = 0.1, 2.0, 1.6, 4
rng = random.Random(51)
print("A  rigidity r   H (bits)   Gini G   pi_s    w     bias at bottom"
      " (g 1.6, L 4)")
rows = []
for r in (0.0, 0.5, 0.8, 0.95, 1.0):
    h, g = cond_entropy(sequence(rng, r, 60000))
    pis = 1 / (S0 + g); w = pis / (pis + PI_P); b = (GAIN * (1 - w)) ** LEVELS
    rows.append((r, w, b))
    print(f"   {r:9.2f} {h:10.3f} {g:9.3f} {pis:7.2f} {w:6.3f} {b:12.3f}")
print(f"   uniform maximum H = log2(6) = {math.log2(ACTIONS):.3f} bits")
w_star = 1 - 1 / GAIN
print(f"B  the bias is damped when g(1-w) < 1, i.e. w > 1 - 1/g = {w_star:.3f}"
      f" (g = {GAIN})")
for g in (1.2, 1.6, 2.0, 3.0):
    print(f"   g = {g:.1f}: threshold w* = {1 - 1 / g:.3f};"
          f" needed pi_s > {PI_P * (1 - 1 / g) / (1 / g):.2f}"
          f" (prior precision {PI_P})")
# C: volatility estimate = running mean of squared prediction errors over two channels,
#    the agent's own actions (Gini as above) and an outside signal with variance 1.
for r in (0.0, 0.95):
    h, g = cond_entropy(sequence(random.Random(7), r, 60000))
    for share in (0.0, 0.5, 0.8):
        v = share * g + (1 - share) * 1.0
        print(f"C  r = {r:.2f}, attention share on own actions {share:.1f}:"
              f" volatility estimate {v:.3f} (world 1.000)")
# A  rigidity r   H (bits)   Gini G   pi_s    w     bias at bottom (g 1.6, L 4)
#         0.00      2.585     0.833    1.07  0.349        1.178
#         0.50      1.954     0.627    1.37  0.407        0.808
#         0.80      1.033     0.299    2.51  0.556        0.254
#         0.95      0.338     0.079    5.59  0.736        0.032
#         1.00      0.000     0.000   10.00  0.833        0.005
#    uniform maximum H = log2(6) = 2.585 bits
# B  the bias is damped when g(1-w) < 1, i.e. w > 1 - 1/g = 0.375 (g = 1.6)
#    g = 1.2: threshold w* = 0.167; needed pi_s > 0.40 (prior precision 2.0)
#    g = 1.6: threshold w* = 0.375; needed pi_s > 1.20 (prior precision 2.0)
#    g = 2.0: threshold w* = 0.500; needed pi_s > 2.00 (prior precision 2.0)
#    g = 3.0: threshold w* = 0.667; needed pi_s > 4.00 (prior precision 2.0)
# C  r = 0.00, attention share on own actions 0.0: volatility estimate 1.000 (world 1.000)
# C  r = 0.00, attention share on own actions 0.5: volatility estimate 0.917 (world 1.000)
# C  r = 0.00, attention share on own actions 0.8: volatility estimate 0.867 (world 1.000)
# C  r = 0.95, attention share on own actions 0.0: volatility estimate 1.000 (world 1.000)
# C  r = 0.95, attention share on own actions 0.5: volatility estimate 0.541 (world 1.000)
# C  r = 0.95, attention share on own actions 0.8: volatility estimate 0.265 (world 1.000)

A fully random sequence (r=0r = 0) has an entropy of 2.585 bits, that is log⁡26\log_2 6; at r=0.95r = 0.95 it falls to 0.338 bits and at r=1r = 1 to 0. With g=1.6g = 1.6, L=4L = 4 and prior precision 2, the random sequence gives a sensory weight w=0.349w = 0.349, below the threshold 0.375, and the bias is amplified to 1.178; at r=0.8r = 0.8, w=0.556w = 0.556 and the bias falls to 0.254; at r=0.95r = 0.95 to 0.032. The threshold depends on the gain: with g=3g = 3 one needs πs>4\pi_s > 4, that is w>0.667w > 0.667. Figure 1 shows both parts.

Left: conditional entropy and Gini impurity of the action sequence against rigidity r. Right: in the author’s cascade model, the bias reaching the bottom level against sensory precision for four gains (log scale; dotted line at 1).

In this model we can say: if sensory precision rises with the predictability of action, and if bias passes and grows from level to level as above, then a sufficiently rigid action sequence pushes the bias below threshold. None of the three “ifs” has been measured. The crucial one is step two: proprioceptive precision is set by the noise of receptors and afferent pathways, not by whether an action follows a script; equating “predictable” with “precise” is a modelling choice, not a physiological fact.

The dark room problem

Part C of the program exposes a structural loophole in the model. Suppose the agent estimates “how uncertain the world is” from a weighted average of two kinds of prediction error: one from its own actions, the other from an outside signal of fixed variance 1. When it puts 80 % of its attention on a rigid sequence with r=0.95r = 0.95, its estimate of the world’s uncertainty falls from 1.000 to 0.265, though nothing in the world has changed; with random actions it falls only to 0.867. In other words, in this model ritual lowers “anxiety” precisely by letting the agent use predictable input it produces itself to underestimate the uncertainty of the outside world.

This is a variant of the famous “dark room problem” of the free-energy literature: if organisms only minimise surprise, why do they not hide in a dark, quiet room? The answer of Friston, Thornton and Clark is that an organism’s priors include expectations of certain sensory states (food, warmth, company), so hiding in a dark room produces more surprise in the long run (Friston et al. 2012). Applied to ritual, the conclusion cuts both ways. A short stretch of rigid action may let a person step away from thoughts of threat for a while, which fits Boyer and Liénard’s “filling working memory”; but when ritual becomes a way of avoiding information from outside, in this model it becomes a mechanism for underestimating real risk. Ritual in obsessive–compulsive disorder is a reminder: rigid, repeated action can lower anxiety for the moment and sustain it over time.

Hypothesis · not established · Rigid action sequences and acute anxiety

The author’s hypothesis (not established): after an acute anxiety induction, completing a 3-minute action sequence in a fixed order lowers self-reported anxiety and heart rate faster than completing a sequence of the same duration, the same kinds of action and the same attentional demand in random order; and the size of the effect rises monotonically with the predictability of the sequence (for example, measured by conditional entropy). What would show it wrong: with a pre-registered sample size and analysis plan, a 95 % confidence interval for the difference in anxiety and heart-rate recovery that lies within the pre-specified smallest effect of interest; or an effect that appears only when participants are told “this is a ritual” and does not depend on the predictability of the sequence (which would show that expectation, not rigidity, is doing the work).

Chapter summary

“Ritual is mere superstition” can be falsified: Lang and colleagues showed that anxiety makes movement more repetitive and rigid (partly replicated in the pre-registered study of Karl and Fischer), and Brooks and colleagues and Norton and Gino showed that ritual is sometimes followed by lower anxiety and better performance, though Karl and Fischer’s pre-registered study found essentially null results in the “stress reduction” direction. But the first finding is “anxiety elicits ritual”, and the second rests on small samples and depends on instructions, with no large pre-registered replication yet; simulation shows that in such conditions published effects may be several times too large. The author’s precision-weighting model states the condition g(1−w)<1g(1-w) < 1 under which, in this model, a rigid sequence damps bias, and it also exposes the dark-room loophole: the same mechanism makes one underestimate the uncertainty of the world. Ritual is more than superstition, and less than a cure-all.

References

Boyer, Pascal, and Pierre Liénard. 2006. “Why Ritualized Behavior? Precaution Systems and Action Parsing in Developmental, Pathological and Cultural Rituals.” Behavioral and Brain Sciences 29 (6): 595–613.
Brooks, Alison Wood, Juliana Schroeder, Jane L. Risen, et al. 2016. “Don’t Stop Believing: Rituals Improve Performance by Decreasing Anxiety.” Organizational Behavior and Human Decision Processes 137: 71–85. https://doi.org/10.1016/j.obhdp.2016.07.004.
Friston, Karl. 2010. “The Free-Energy Principle: A Unified Brain Theory?” Nature Reviews Neuroscience 11 (2): 127–38. https://doi.org/10.1038/nrn2787.
Friston, Karl, Christopher Thornton, and Andy Clark. 2012. “Free-Energy Minimization and the Dark-Room Problem.” Frontiers in Psychology 3: 130. https://doi.org/10.3389/fpsyg.2012.00130.
Hobson, Nicholas M., Juliana Schroeder, Jane L. Risen, Dimitris Xygalatas, and Michael Inzlicht. 2018. “The Psychology of Rituals: An Integrative Review and Process-Based Framework.” Personality and Social Psychology Review 22 (3): 260–84.
Karl, Johannes A., and Ronald Fischer. 2018. “Rituals, Repetitiveness and Cognitive Load: A Competitive Test of Ritual Benefits for Stress.” Human Nature 29: 418–41.
Lang, Martin, Jan Krátký, John H. Shaver, Danijela Jerotijević, and Dimitris Xygalatas. 2015. “Effects of Anxiety on Spontaneous Ritualized Behavior.” Current Biology 25 (14): 1892–97. https://doi.org/10.1016/j.cub.2015.05.049.
Norton, Michael I., and Francesca Gino. 2014. “Rituals Alleviate Grieving for Loved Ones, Lovers, and Lotteries.” Journal of Experimental Psychology: General 143 (1): 266–72.

Fan, Y. (2026). Neurobiology of Ritual, Spatial Boundary Transitions and Computational Biomedicine of Self-Transcendence: From Predictive-Coding Free-Energy Minimisation, Default-Mode Network Decoupling and Vagal–Endocrine Restoration to Classical Chinese Medicine on Shifting the Spirit (Yijing Bianqi) [仪式神经学、空间边界切换与自我转移的计算生物医学] (Version 1.0.0). 5D Library. https://doi.org/10.5281/zenodo.23839910 · Licensed under Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International (CC BY-NC-ND 4.0).