Depth of Processing and the Self-Reference EffectMany experiments have been done on the depth of processing and the self-reference effect. The depth-of-processing model of memory holds that the depth with which something is encoded in a person's memory depends on the use of certain types of processing. This refers to the self-reference effect because it is believed that people have a tendency to remember something better when they can relate it to themselves. People who can personally relate to something have a tendency to ingrain it more deeply in their memory. Craik and Tulving did a series of experiments on the depth-of-processing model. They asked participants to use a variety of processing methods to encode words at different levels; superficial, moderate and deep. Subjects were shown a series of words and asked questions about the words that would provide a “yes” or “no” answer. On a superficial level they were asked questions about whether or not the word was capitalized. At the moderate level of elaboration, the subject was asked whether or not two words rhymed. Finally, the subjects were asked which words were present in the sentences and whether they were suitable or not. This was the deep level of processing. After participants completed the task, they were given a surprise recognition test with the words they had been asked questions about (target words) and then words they had never seen before (distractor words). The results of the experiment showed that people remembered words better that were at a deeper level of processing (Craik and Tulving 1975). Although there has been some criticism of the above experiment, Craik and Tulving each time performed more experiments refining the PDO model. It was thought that the structural tasks were easier and that you didn't have to spend as much time on them, so people didn't have as much time to look at those words and couldn't study them like the other tasks. Craik and Tulving then made the structural task take the same amount of time as the other tasks. The results remain the same as previous experiments. Craik and Tulving also initially started with five tasks, but then narrowed it down to three to avoid a ceiling effect. The self-referential task was later added to the model by Rogers.