Twenty Thousand Leagues: Depth of Processing

Learning By Doing

Without cheating, take a moment to scan the following matrix of pennies. 

Which drawing represents the current design of a penny? (Fun fact: the front-side of a coin is called the obverse.)

Figure 1. Can the real penny please stand up?

Once you've made your selection, reflect on the following questions:

1. Was this a difficult task?
2. If so, why was this difficult?
3. If not, what life experience did you have that prepared you to answer quickly and accurately?

Like a Bad Penny...

I've used the penny demonstration in several groups over the past year, and I would estimate that about 1 in 30 select the right coin [1]; therefore, it's likely this task was hard for you, too. The question is why (despite the fact we've all seen a penny many, many times)? 

One reason is related to the first of three processes in our simple model of memory: encoding. You might be unable to identify the real penny, in a sea of distractors, because you never encoded the specific properties of a penny. 

On the other hand, you probably did encode the approximate size, the edge's smoothness, the inclusion of our 16th president, and the copper veneer. You encoded those properties because they help you meet your learning goal: to spend a penny. You need to differentiate a penny from a nickel, dime, and quarter. In other words, learning is goal oriented. If your goal changes (e.g., you need to accurately draw a penny for your blog), then you will encode those specific details.

That's Deep.

Misidentifying a penny among 14 distractors is a pretty powerful example of encoding failure. We've all seen a penny before, but it's unlikely that someone quizzed you on the specific features. So what if they did? Is there a way to enhance our ability to remember something over the long haul? In other words, is all processing is equally effective? I'm sure the answer is obvious to you. But what's the evidence?

To test if different levels of processing lead to better (or worse) memories, two very famous memory researchers, Fergus Craik and Endel Tulving, conducted a series of experiments [2]. They asked volunteers to come to their lab and study a list of words. They experimentally manipulated how deeply the participants studied the words across five levels of processing. They asked the participants (1) if the word was present; (2) if the word was written in capital letters; (3) if the word rhymed with another word; (4) if it was in a specific category;  or (5) if it fit into a given sentence. Table 1 contains both positive and negative examples for the word honey.

Table 1. The experimental manipulation of processing depth. 

Level of Processing	Positive Example	Negative Example
1. Is there a word present?	HONEY
2. Is the word in capital letters?	HONEY	honey
3. Does the word rhyme with ___?	funny	monk
4. Is the word in a category of ___?	a type of food	a part of a ship
5. Does the word fit in this sentence: ___?	He eats ______ on his toast.	She met a ____ on the street.

Before you look at the results, what do you predict they found? Does processing a word at a deeper, more semantic level, lead to better memory trace? Or is processing all the same, as long as it passes through the attentional system and into working memory? 

Here's what they found: 

Figure 2. Memory accuracy as a function of depth of processing.

According to their results, there is a huge advantage for processing a word at a deeper, semantic level. 

The Classroom Connection

The evidence from Craik and Tulving's study is highly suggestive. As educators, we want our students to process the information we want them to learn as deeply as possible. This advice is consistent with other research on memory:

1. We saw evidence that generating a to-be-remembered item can also have beneficial effects on long-term recall. 
2. Research on retrieval practice demonstrated that rereading notes or a chapter summary was a terrible way to study. Instead, students should engage in an effortful attempt to retrieve the information from memory. 
3. We also reviewed a study that showed students prefer easy hints to more difficult hints when self-testing. Unfortunately, their preferences ran completely counter to the empirical evidence suggesting that more effortful retrieval leads to better retention. 

Because encoding is a vitally important step for robust learning, it is helpful to understand what the empirical data has to say: The deeper the processing, the longer lasting the memory. 

Process that as deeply as you can! 🦑

Share and Enjoy!

Dr. Bob

Going Beyond the Information Given

[1] By the way, the right answer is (a). Don't believe me? I bet the US Mint will back me up.

[2] Craik, F. I., & Tulving, E. (1975). Depth of processing and the retention of words in episodic memory. Journal of experimental Psychology: General, 104(3), 268.

The Thin Red Line: Precise Elaboration

Below are two lists of sentences. The first list contains short sentences with the following structure: <definite article> <adjective> <subject> <verb> <article> <direct_object>. The second list contains long sentences with the same structure, plus an extra prepositional phrase (in italics) at the end that modifies the base sentence. 

After reading the sentences, suppose I were to give you a memory test. Which sentences do you think would be easier for you to remember? Why is that the case?

Short Sentences

1. The short man bought the broom. 
2. The old man used the paint.

3. The fat man looked at the warning.

Long Sentences

1. The short man bought the broom to operate the light switch.

2. The old man used the paint to change the color of his cane.

3. The fat man looked at the warning that said keep off the thin ice.

Many of the previous posts about memory talked about memorizing individual items, like numbers, letters, or words. While it is essential to be able to remember small bits of information, it is also important that we have the ability to remember longer, more complicated pieces of information. Reading often requires the ability to encode large blocks of complicated information. What kind of strategies exist to help boost our memory for sentences? 

Elaborative Processing

As stated previously, one strategy to boost your memory is to add supplemental information. It seems counter-intuitive that adding information would increase the likelihood of remembering because now you have more to remember! It works because the mind craves two things: order and redundancy.

To demonstrate, a pair of scientists constructed some cleverly-worded sentences and gave them to participants to read [1]. The first type of sentence, or base sentences, included two nouns connected by a verb. Here is an example: 

Base Sentence: The woman hit the butcher. 

For the second type of sentence, the scientists embellished the base sentences with some color commentary, like this:

Embellished Sentence: The woman hit the butcher with a sausage.

Half of the participants read the base sentences and the other half read the embellished sentences. After they studied the sentences, the experimenters then checked the accuracy of the participants' memory for the underlined words. You can probably guess the outcome of the experiment. If you guessed that the people who read the base sentences remembered fewer items (57%) than those who read the embellished sentences (72%), then you would be correct. The explanation was that the additional information helped create a rich context for remembering the other words in the sentence. The additional commentary helped glue everything together. 

"Doctor, isn’t that incision a bit high for an appendix?"

This is an interesting finding because it shows that our memories are sensitive to the information that surrounds the target information. However, the results left a couple of open questions. First, does the type of elaboration matter? By type I am referring to the relevance of the elaboration to the main sentence. An imprecise elaboration does not add relevant information. On the other hand, a precise elaboration adds information that is semantically connected to the base sentence. A precise elaboration, for example, might explain why being "thin" is relevant to the developing story. 

The second open question deals with the generation effect, which distinguishes between elaborations that are provided by someone else or that we generate ourselves. In the experiment described in the previous section on elaborative processing, the elaborations were provided by the experimenters. What if people were able to supply their own color commentary? Would that prove to enhance memory even more?

To test these questions, some other scientists wrote another set of cleverly-worded sentences that the they gave participants to study [2]. The sentences included the following four types:

1. Base sentence: The thin man picked up the scissors.
2. Imprecise-elaboration: The thin man picked up the scissors to cut the tag off his hat. 
3. Precise-elaboration: The thin man picked up the scissors to cut the belt in half.
4. Self-elaboration: The thin man picked up the scissors ______________.

There were two phases to the experiment. For the first phase, participants either read the sentences that they were given (base, imprecise-elaboration, or precise-elaboration sentences), or they read the base sentence and generated their own elaboration (self-elaboration). During the second phase of the experiment, participants were given the base sentence with the adjective removed (e.g., thin), and their job was to recall the missing word.

Before I describe the results, can you guess the order of the conditions in terms of their performance on the cued recall test?

Fig. 1: Results from the elaboration study.

As you can see from the graph, the individuals who read sentences that were imprecisely elaborated performed the worst. They recalled fewer adjectives than the participants in the other conditions of the experiment. The best performance was found in the group that read the precisely elaborated sentences, with the self-elaboration not too far behind. 

In summary, I guess it would be a mistake to say that merely adding additional information helps boost our memory. Instead, the additional information needs to be semantically relevant.

The STEM Connection

The research on elaboration has useful applications for education. First, textbooks should be written with the second study in mind. That is, textbook authors should strive to ensure that their elaborations are precisely worded and relevant to the material being presented. The goal is to create a coherent mental model for the reader. For example, it would be insufficient to say, "Hot air rises." because an explanatory mechanism is not mentioned. Instead, the reader is forced either to memorize this bit of information or supply their own explanation for why hot air rises. Self-generated elaborations aren't necessarily a bad thing (as we will see in a future post), but the reader might be lazy and neglect the intellectual work needed to understand the passage.

The second application is on the student side. Students should train themselves to supply their own explanatory or causal mechanism when it is missing. If a piece of text doesn't make sense or seems incomplete, the reader should ask, "Why does hot air rise?" This type of self-prompting is beneficial because the reader is effectively training herself to become more meta-cognitively aware.

The research behind elaborative processing and precisely worded elaborations is interesting because we've started to move past merely memorizing lists of words; however, we are still at the sentence level. The next goal is to figure out how we process larger chunks of texts – like entire paragraphs! 


Share and Enjoy!

Dr. Bob

For More Information

[1] Anderson, J. R., & Bower, G. H. (2014). Human associative memory. Psychology Press.

[2] Stein, B. S., & Bransford, J. D. (1979). Constraints on effective elaboration: Effects of precision and subject generation. Journal of Verbal Learning and Verbal Behavior, 18(6), 769-777.