The demonstration that color perception can be modified through learning indicates that the process of learning a language may influence or even shape the way people perceive colors. However, observed differences between speakers of different languages when they perform various cognitive tasks such as similarity judgments and even visual search do not provide sufficient evidence that this is actually what happens. As mentioned earlier, these differences might be attributable to differences in verbal labeling as opposed to pure perception. Therefore, it is necessary to carry out studies focusing on low-level perception of color by speakers of different languages. For example, the color-difference detection thresholds of speakers of languages with different color-term repertoires could be compared. Any observed differences in thresholds for an area that is around a linguistic boundary in one language but not the other would most likely reflect differences in low-level perceptual sensitivities. Such investigations are currently under way in our laboratory. Demonstrating such low-level differences between speakers of different languages using this or similar methods would be compelling evidence for the strong form of the linguistic relativity hypothesis. Another issue that needs to be resolved concerns infants’ color perception. As mentioned, after habituating to a color, 4-month-old infants seem to prefer looking at a color from a different universal category over looking at another color from the habituated category. This suggests that the universal categories may be ‘‘hardwired.’’ However, on some tasks, school-age African children show evidence that their own language influences their perception of a color (a linguistic relativity effect). The most compelling way of resolving these apparently conflicting findings might be to use direct measures of perceptual sensitivity (such as color-difference detection thresholds), adapted for preschool children or even infants. Infant studies of color perception have mainly used ‘‘preferential looking’’ (i.e., amount of time spent looking at one stimulus compared with another) as a measure. However, it is difficult to know exactly what perceptual ability this measure taps into; it is conceivable that infants are able to discriminate between many colors but ‘‘prefer’’ to look at certain stimuli rather than others for a reason that is not related to perceptual sensitivity. A measure suitable for assessing infants’ ability to discriminate colors would be likely to provide crucial insight into the origin of color CP. It is remarkable to think that even the most fundamental perceptual mechanisms, such as those involved in the perception of colors, can be influenced by the environment, including culture and language. The evidence reviewed here suggests just such an influence. So, to go back to my initial example, it is just possible that what you see when you look at the rainbow actually depends on the language you speak