SweetPea Window Examples

sweetpea-window-examples.py

# SweetPea Window Examples

# Derived Levels are defined by three things:
# - A Derivation Function
# - An Argument List
# - A Window
#
# Windows in turn are defined by a width and a stride. WithinTrial and
# Transition are just shorthand for Windows with widths of 1 and 2
# respectively, and both with a stride of 1.
#
# Here's an example of a Window with width 1 and stride 1:
con_factor = Factor("congruent?", [
    DerivedLevel("con", WithinTrial(lambda color, text: color == text, [color, text])),
    DerivedLevel("inc", WithinTrial(lambda color, text: color != text, [color, text]))
])

# And here's an example of a Window with width 2 and stride 1:
repeated_color_factor = Factor("color repeats?", [
    DerivedLevel("yes", Transition(lambda colors: colors[0] == colors[1], [color])),
    DerivedLevel("no",  Transition(lambda colors: colors[0] != colors[1], [color]))
])

# Notice how the derivation function for the repeated color factor receives
# a _list_ of color values. (colors[0], colors[1], etc)
#
# In contrast, notice how the congruent factor receives just a color and text
# value directly.
#
# For all Windows with a width of 1, the derivation function receives the value
# for each factor directly, rather than a list, because there is only one value
# when the width is 1.
#
# On the other hand, when the width is greater than 1, the derivation function
# receives a _list_ of values for each factor, where the number of values for
# each factor corresponds to the width of the Window.

# Here's a slightly more complicated example:
congruent_bookend = Factor("congruent bookend?", [
    DerivedLevel("yes", Window(lambda color, text: color == text, [color, text], 1, 3)),
    DerivedLevel("no",  Window(lambda color, text: color != text, [color, text], 1, 3))
])

# This Window has a width of 1, and a stride of 3. Because the width is 1, the
# function receives factor values directly, not lists. This factor takes the
# 'yes' level whenever every third trial (starting with the first) is congruent.

# Consider another example, where we want to identify whether or not the first
# and third trials in a sliding window have the same level. We can define such
# a factor like this:
first_third_match = Factor("first and third match?", [
    DerivedLevel("yes", Window(lambda colors: colors[0] == colors[2], [color], 3, 1)),
    DerivedLevel("no",  Window(lambda colors: colors[0] != colors[2], [color], 3, 1))
])

# This window has a width of 3, and a stride of 1. Notice again how, because
# the width is greater than 1, the derivation function receives a _list_ of 
# values for each factor. The nth value in the list corresponds to the factor's
# value in the nth trial of the Window.