i.e. is it a convention when publishing a priming experiment to use different cases to distinguish between prime and target, or are different cases used in the actual experimental scripts using ePrime or DMDX. If the latter, what's the rationale?
Thanks Stephen, that's a very interesting point about the pixel-level match. Would the same rationale hold for semantic/associative priming? I ask because I came across a passage in Alderson (2000:75) Assessing Reading which said that upper-case letters are harder to process than lower-case and so, I infer (!), would affect reaction times.
Dear Michael,
priming can happen on multiple levels: physical overlap (e.g., pixels: hunt - hunt), formal priming at the orthographic and phonological level (corn - corner, table - able), at the morphological level (teach - teacher), at the semantic level (far - distant), at the syntactic level (the - table, versus to - table) and so on. While response times in behavioural studies may not always differentiate between them, event-related brain potentials (ERPs) have identified distinct brain waves (reflecting different sub-processes) for at least some of these levels. As Stephen already pointed out, pixel overlap is one issue, but retinal after-images is another one. If you include a condition of identical priming (hunt - hunt) to establish a condition with maximal priming (often useful to determine how strong other primes are relative to maximal priming), then the eye may not even notice that there were two words on the screen (the prime 'hunt' and then the target 'hunt'), especially if your ISIs are short. In this case it's absolutely crucial to have different fonts, or upper-case vs lower-case letters for prime and target, respectively. In my lab, we typically create different versions of the same experiment (also in terms of randomizations), where the order of upper and lower case letters is balanced between prime and targets as well as across subjects (e.g., one subject would see prime1 - TARGET1, PRIME2 - target2, and another subject would see PRIME 1 - target1, prime2 - TARGET2, and then in inverse orders for subjects 3 and 4). I believe this addresses the issues of overlap and sequence effects best (and also allows you to test if upper and lower case words differ in processing time, etc). Another effective way to completely erase retinal after images is the use of masked priming. Please see attached file of and ERP study for more details on all issues discussed above.
I hope this helps,
Karsten
- Badges
- Science topic
- Similar topics
- Mathematics