Linda
Liu

Ideal Categorizer Demo

Florian Jaeger and I recently put together an R shiny app that simulates an ideal categorizer (following the ideal observer, in the Bayesian sense) for two sound contrasts (e.g., /p/ and /b/).

Voice onset time (the time between the release of a stop consonant and when the vocal folds start vibrating) is one of major cues in determining whether a sound is a /p/ or a /b/ sound. By manipulating its value, we can get sounds on a continuum from /b/ to /p/.

Here you can manipulate the category prior, along with the mean and standard deviation for each sound category (left side). On the right side, the top panel shows the likelihood function as determined by those means and standard deviations. The middle panel shows the posterior -- the likelihood combined with the category priors. Finally, the bottom panel predicts what we might expect the categorization response curve to look like based on this posterior.

Amazon Mechanical Turk + Boto Tutorial (Part I)

This is Part I in a series on how to crowdsource your tasks/experiments on Amazon Mechanical Turk. In this handout, I describe how to set up an Amazon Mechanical Turk account, link it to boto (python package for interacting with MTurk), and obtain results from a (premade) HIT hosted on an exteneral server using some (provided) python scripts.

I originally created this handout as part of an assignment for a graduate-level data science course I co-taught with Florian Jaeger.

Amazon Mechanical Turk (Part II)

Here I go over how to create your own external HIT (based on a previous experiment by Kleinschmidt and Jaeger 2012). Note that you will need your own web server in order to run an external HIT.

I originally created this handout as part of an assignment for a graduate-level data science course I co-taught with Florian Jaeger.

PixelPy - AOI labeling tool for visual stimuli

PixelPy makes it easier to label areas of interest in visual scenes. This is especially relevant for eyetracking experiments that involve more complex visual scenes. See the linked github page to download and for sample usage.