Data for "Why Bananas Look Yellow: The Dominant Hue of Object Colours"

These extended supplementary materials go with the article:

• Witzel & Dewis (2022) Why Bananas Look Yellow: The Dominant Hue of Object Colours. Vision Research.

A. SURVEYS

A pdf-printout for each of the three Qualtrics surveys illustrates details of the procedure. The layout may have been slightly different in Qualtrics (e.g., wide screen vs portrait display). Also note that the second and third surveys feature a few questions that were unrelated to the dominant-hue study (identifying a grey image).

B. STIMULI

The images used in Experiments 1-3, and the animated images used as cues to colour changes in Experiment 3 are packed in zip-files.

C. CODE

The Matlab code "onehue_maker.m" is a function that implements the dominant-hue algorithm to produce one-hue images like those in the experiments. To try out the program, the photo of the banana and the mask identifying its background are also uploaded (= first and second input to the function). The purpose of the mask is to remove the background colour from the dominant-hue computations.   

D. DATA

The uploaded data is not completely raw but has been polished in the following ways:

• Pilot data has been removed (i.e., meaningless data from us and our students to try out, check and polish the survey).
• Incomplete runs have been removed (i.e., when participants quitted before completing the whole survey).
• Data irrelevant to this study have been removed (date and time; grey-identification task [see above]).

There are 3 sheets with data and three sheets with stimulus specifications for each of the three experiments. The stimulus specifications include the measures used in the analyses in "Other Factors" in the Discussion of Experiment 3. 

Columns in the Data sheets are:

• Participant information: recruit (soc med = social media; UG pool = undergraduate students, prolific = https://www.prolific.co/); coldef = Colour deficiencies (1 Yes, 2 No according to test, 3 No without test, 4 Don't know); sex (1 male, 2 female, 3 other); age (in years), and duration (in minutes).
• Main data: Column labels are composed of the following elements, separated by an underscore (_):
  • The first 3-5 letters of the object name: ban = banana, car = carrot, cher = cherry, dress = #theDress, fro = frog, gra = grapes, lem = lemon, let = lettuce, ora = orange, pig, ros = rose, shoe = #theShoe, stra = strawberry, zuc = zucchini/courgette.
  • A symbol indicating the stimulus condition: 1 = One-Hue, m = Minus-Hue Rotation, p = Plus-Hue Rotation.
  • A number identifying the measure: 1 = responded position; 2 = accuracy of the response (1 = correct); 3 = response time (in sec), 4 (Experiment 2-3) = confidence rating (between 0 and 100), 5 (Experiment 3) = cue confidence (cf. Figure 11.a).
  • For inverted colours (Experiment 3), the column label starts with an "i" (for inverted).
• Practice Trials: Start with the prefix ex (for example) followed by an underscore (_) and the ID of the object; otherwise, data as in main trials.
• Catch Trials (Experiment 2-3): Start with object name "d" for disk, otherwise, data as in main trials. 
• Eidolon Guesses (Experiment 2): Start with "guess" followed by the object ID (see main trials) followed by a number indicating the measure: 1 = response (yes/no), 2 = confidence (if positive response). In case of a positive response, the text entries are save in the variables starting with guess_txt.

Columns in the stimulus sheets are:

• DomHue: Angle of the dominant hue (cf. Figure 3); as principal components are relative to the average, the angle is relative to the average, not the origin.
• pole1 and pole2: Poles of the dominant hue direction. "pole1_rgb" provides corresponding RGBs for illustration (cf. Figure 1).
• ChromaRescaled: Rescale Factor (see Experiment 3).
• MaxChr: Maximum chroma of the colour distribution in CIELUV.
• M: Average chromaticities (u*, v*) of the colour distribution.
• pc: Coefficients of the first principal component for u* and v*.
• latent & expl: Absolute and relative explained variance, respectively; second column corresponds to orthogonal variance.
• hueM & hueSD: Average and standard deviation of the hue of the colour distribution (cf. Figure 3).
• rot_minus, rot_plus: The hue rotations in the rotated-hue condition (constant minus or plus 5, except for #theShoe).
• oog_1hue, oog_plus, oog_minus: The proportion of out-of-gamut values.
• oogdist_1hue, oogdist_minus, oogdist_plut: Average difference between clipped and original images (in CIELUV).
• Mshift_1hue, Mshift_minus, Mshift_plus: Average and standard deviation of chromaticity shift due to the experimental manipulation (cf. Figure 5 and Table S1).
• Mhueshift_1hue, Mhueshift_minus, Mhueshift_plus: Average and standard deviation of hue shift in CIELUV (cf. Figure S4.d-f and Table S2).
• Lab_shift_1hue, Lab_shift_minus, Lab_shift_plus: Average and standard deviation of chromaticity shift in CIELAB (cf. Figure S4.a-c and Table S1).
• Lab_hueshift_1hue, Lab_hueshift_minus, Lab_hueshift_plus: Average and standard deviation of hue shift in CIELAB (cf. Figure S4.g-i and Table S2).
• Lab_Mhue: Hue of the average colour in CIELAB
• Lab_hueM & Lab_hueSD0: Average and standard deviation of the CIELAB hue distribution.
• huehist0: CIELUV hue histogram; each entry corresponds to the frequencies for 72 bins of 5-deg (cf. Figure 3); the zero indicates that the hue is relative to the origin, not to the average chromaticity.