Dr Yun-Xiao Li presenting a poster at a conference
Welcome to Dr Yun-Xiao Li, who will be investigating how we make choices about risk and the future in his ESRC-funded postdoc with Dr. Manos Konstantinidis. Yun-Xiao has not travelled far: he did his PhD in our department on discreteness in mental sampling and decision making.
![R code and output showing the new functionality:
``` r
## pak::pkg_install("quentingronau/bridgesampling#44")
## see: https://cran.r-project.org/web/packages/bridgesampling/vignettes/bridgesampling_example_stan.html
library(bridgesampling)
### generate data ###
set.seed(12345)
mu <- 0
tau2 <- 0.5
sigma2 <- 1
n <- 20
theta <- rnorm(n, mu, sqrt(tau2))
y <- rnorm(n, theta, sqrt(sigma2))
### set prior parameters ###
mu0 <- 0
tau20 <- 1
alpha <- 1
beta <- 1
stancodeH0 <- 'data {
int<lower=1> n; // number of observations
vector[n] y; // observations
real<lower=0> alpha;
real<lower=0> beta;
real<lower=0> sigma2;
}
parameters {
real<lower=0> tau2; // group-level variance
vector[n] theta; // participant effects
}
model {
target += inv_gamma_lpdf(tau2 | alpha, beta);
target += normal_lpdf(theta | 0, sqrt(tau2));
target += normal_lpdf(y | theta, sqrt(sigma2));
}
'
tf <- withr::local_tempfile(fileext = ".stan")
writeLines(stancodeH0, tf)
mod <- cmdstanr::cmdstan_model(tf, quiet = TRUE, force_recompile = TRUE)
fitH0 <- mod$sample(
data = list(y = y, n = n,
alpha = alpha,
beta = beta,
sigma2 = sigma2),
seed = 202,
chains = 4,
parallel_chains = 4,
iter_warmup = 1000,
iter_sampling = 50000,
refresh = 0
)
#> Running MCMC with 4 parallel chains...
#>
#> Chain 3 finished in 0.8 seconds.
#> Chain 2 finished in 0.8 seconds.
#> Chain 4 finished in 0.8 seconds.
#> Chain 1 finished in 1.1 seconds.
#>
#> All 4 chains finished successfully.
#> Mean chain execution time: 0.9 seconds.
#> Total execution time: 1.2 seconds.
H0.bridge <- bridge_sampler(fitH0, silent = TRUE)
print(H0.bridge)
#> Bridge sampling estimate of the log marginal likelihood: -37.73301
#> Estimate obtained in 8 iteration(s) via method "normal".
#### Expected output:
## Bridge sampling estimate of the log marginal likelihood: -37.53183
## Estimate obtained in 5 iteration(s) via method "normal".
```](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:lgxrkw556dtxkplwrj5rzlli/bafkreicf6zdyyjmaxrdfyq2sekokprym3c2o5atca6i2cczcx5zec4q4hm)
R code and output showing the new functionality: ``` r ## pak::pkg_install("quentingronau/bridgesampling#44") ## see: https://cran.r-project.org/web/packages/bridgesampling/vignettes/bridgesampling_example_stan.html library(bridgesampling) ### generate data ### set.seed(12345) mu <- 0 tau2 <- 0.5 sigma2 <- 1 n <- 20 theta <- rnorm(n, mu, sqrt(tau2)) y <- rnorm(n, theta, sqrt(sigma2)) ### set prior parameters ### mu0 <- 0 tau20 <- 1 alpha <- 1 beta <- 1 stancodeH0 <- 'data { int<lower=1> n; // number of observations vector[n] y; // observations real<lower=0> alpha; real<lower=0> beta; real<lower=0> sigma2; } parameters { real<lower=0> tau2; // group-level variance vector[n] theta; // participant effects } model { target += inv_gamma_lpdf(tau2 | alpha, beta); target += normal_lpdf(theta | 0, sqrt(tau2)); target += normal_lpdf(y | theta, sqrt(sigma2)); } ' tf <- withr::local_tempfile(fileext = ".stan") writeLines(stancodeH0, tf) mod <- cmdstanr::cmdstan_model(tf, quiet = TRUE, force_recompile = TRUE) fitH0 <- mod$sample( data = list(y = y, n = n, alpha = alpha, beta = beta, sigma2 = sigma2), seed = 202, chains = 4, parallel_chains = 4, iter_warmup = 1000, iter_sampling = 50000, refresh = 0 ) #> Running MCMC with 4 parallel chains... #> #> Chain 3 finished in 0.8 seconds. #> Chain 2 finished in 0.8 seconds. #> Chain 4 finished in 0.8 seconds. #> Chain 1 finished in 1.1 seconds. #> #> All 4 chains finished successfully. #> Mean chain execution time: 0.9 seconds. #> Total execution time: 1.2 seconds. H0.bridge <- bridge_sampler(fitH0, silent = TRUE) print(H0.bridge) #> Bridge sampling estimate of the log marginal likelihood: -37.73301 #> Estimate obtained in 8 iteration(s) via method "normal". #### Expected output: ## Bridge sampling estimate of the log marginal likelihood: -37.53183 ## Estimate obtained in 5 iteration(s) via method "normal". ```
Exciting #rstats news for Bayesian model comparison: bridgesampling is finally ready to support cmdstanr, see screenshot. Help us by installing the development version of bridgesampling and letting us know if it works for your model(s): pak::pkg_install("quentingronau/bridgesampling#44")
#rstats #statistics
I've released my new open source book, "Powered by Linear Algebra: the role of matrices and vector space in data science," at matloff.github.io/WackyLinearA....
Turns the classic LA course on its head! Still proves the theorems, but with a deep emphasis on applications.
Our lab has a list of papers that use statistical sampling algorithms like MCMC to explain human behaviour. Thanks to @lcastillo.bsky.social, you can select by behaviour or algorithm.
If we've missed any, please let us know!
sampling.warwick.ac....
Had a great time working with Lucas and Adam on this preprint and the package! Let us know what you think!
Thrilled to see my first publication out in PB&R with @sotarokita.bsky.social and @suzanneaussems.bsky.social! Adults interpret novel verbs using iconic speed cues in speech prosody and hand gesture, with a small but reliable link across modalities.
Open access: link.springer.com/10.3758/s134...
Congrats!๐
5 stars is better than 4 stars, but can we even define how much better it might be? Modeling ordinal outcomes like ratings is a subtle topic; fortunately I have a new chapter that dives directly into that nuance.
HTML: betanalpha.github.io/assets/chapt...
PDF: betanalpha.github.io/assets/chapt...
As a Chinese person, itโs so exciting to see people from outside China engaging with the stories in Zhuang Zi! Your drawings are such a creative interpretationโthank you for sharing this!
New paper with Tong Liu and Arndt Broeder, just accepted in Cognition. We test novel qualitative predictions from sampling-based models of probability estimation in an event ranking task. Results provide evidence for the idea that mental sampling underlies probability judgements.
"Noise in Cognition: Bug or Feature?" is now available in Perspectives on Psychological Science
doi.org/10.1177/1745... (1/4)
(7/7) I want to express my gratitude to Johanna K. Falbรฉn, @lcastillo.bsky.social, Jake Spicer, Jian-Qiao Zhu, Cheng Stella Qian, Nick Chater, and @asanborn.bsky.social for their irreplaceable contribution to this project!
(6/7) In summary, our results suggest:
- ๐๏ธ Adjustment for the assumption of uniform distortions by utility in probability representations;
- ๐ฏ Explicit repeated mental simulation could promote more accurate probability assessments in everyday life.
(5/7) We analysed the data at both the group and individual levels and found:
- Group level: People are optimistic;
- Individual level: The majority remained unbiased; ๐คฏ
- People are optimistic if they can only do the task once;
- People start from a smaller or more probable value if repeated.
(4/7) We manipulated the domains (gain, loss, or neutral), the outcome distributions (equally or unequally distributed), the numbers of outcomes (two, six, or eleven), the repetition of the task (one-off or repeatedly), and the task instructions (imagine, predict, judge the probability).
(3/7) To investigate the mental simulation process, we employed a method called random generation. In this method, participants are asked to imagine playing a gamble and then utter the imagined outcome.
(2/7) People often simulate uncertain events in their heads to assess risk, but do the values of the possible outcomes distort these simulations? Prior experimental findings offer conflicting predictions about how utility may bias this mental sampling process.
๐จ A new preprint is out!
How does utility influence mental simulations of risky events? ๐ค๐ฒ
We tested this across 4 experiments & found that most people simulate probabilities accurately, but biases emerge in key conditions!
If you want to learn more, keep reading!
doi.org/10.31234/osf...
New Preprint Out! ๐๐
Can people generate a random sequence if given enough time?
Keep reading if
- You make cognitive models with randomness in them
- You like to explore the world, be creative, choose well
- You want protection from clever agents exploiting patterns in your behavior.
osf.io/awg9j
