We were keen to investigate the extent to which there would be semantic adaptation. When a normal fluent English speaker hears “red” as part of a sentence, a host of cognitive and behavioral states and processes are invoked that are keyed to a certain phenomenal appearance. If I ask you to “hand me the red block” you can immediately and without overt reflection grasp the correct block, even if it is surrounded by other, differently colored blocks. To what extent would color-rotated subjects show semantic adaptation? Would a point be reached at which the sentence “hand me the red block” just as fluently resulted in the grasp of the block that was in fact red, but presented as blue?
To help facilitate semantic adaptation both subjects spent a good deal of time each day performing tasks requiring engagement with color vocabulary. The most relevant of which was the building game, in which subjects were given written or verbal descriptions for building constructions from colored blocks. The constructions required blocks of specific colors to be placed in specific locations and orientations. Instructions were given in terms of the blocks’ actual color. Success required that subjects select the correct blocks even though the color terms used mismatched their visual input.
Our main method of testing adaptation in a detailed way was Stroop (1935). For technical reasons we have data only for JK (RG’s trail exposed a problem with the interaction between the goggles and the computer display presenting the stimuli, which rendered those data unusable but did allow us to fix the issue so that we could collect valid data from JK). In one standard Stroop set-up, subjects are shown color words presented in colored text, the text either spelling out a color name, or being a neutral series of asterisks, such as ****. For example, subjects might see the word RED in blue text. The task is to name, as quickly as possible, the color of the text while ignoring the color named by the word. When presented with the word RED in blue text, the subject is to say “blue” as quickly as possible.
The standard result is that there is significant interference. Subjects are fastest and most accurate when the color of the text and the color word match, as in the word BLUE in blue text. When there is a mismatch, they are slower and commit more errors, especially errors in which the subject replies with the color word named by the text, and not the color of the text. Our hypothesis was this: with colors rotated by 120° but before a period of adaptation, subjects would show the same pattern as normal subjects in that they would have interference when the color named by the text mismatched the rotated color of the text. So for example the word RED in red text (which would be presented as blue through the goggles) would result in interference, but the word RED in green text (which would be presented as red) would not. But after wearing the gear for a period of time there would be some degree of semantic adaptation. This would manifest in two ways. First, there should be facilitation, or at least diminished interference, when the color named is the same as the actual color of the text, even though it is presented in a different color. And second, there should now be interference, or less facilitation, when the color named is the same as the color in which the word is presented, because that color would be different to the color the word actually is.
The results were that we found no significant effect in either direction. This was disappointing, but it is consistent with the subjective reports of both RG and JK. They both remarked that it quickly became easy to do the appropriate translation and, for instance, grab the green block when one was instructed to “grab the green block” despite its being presented as red through the goggles. But even near the end of their adaptation periods, it still felt like it was an active (though fast and easy) translation, and not a pre-reflective semantic connection between “green” and the actually green objects.