For people who have been teaching for a while the results of this new study probably won’t come as a surprise. I even can imagine that people well familiar with things such as dual coding will have guessed the answer to the question in the title: first the object before the label!
The research, published in Cognitive Science, adds to the existing evidence that adults, children and students of all ages learn better when seeing an object before hearing its description. The study builds on past research by focusing on learning in “inconsistent” environments featuring different teaching styles or distracting noises.
“Understanding how the learning process occurs, and what factors affect it, may help instructors improve methodologies of education,” said Timmy Ma, a research associate at Dartmouth.
Learning environments can often complicate the learning process. For example, a student taking a course with both a teacher and a teaching assistant needs to adapt to the ways the different instructors teach the same subject. Even the varying ways teachers talk and behave can complicate learning.
For the study, researchers intentionally provided confusing information to mimic these types of inconsistencies to subjects that were tasked to learn the names of three fictional characters – “yosh,” “wug” and “niz” – using two types of learning methods.
The first method, “object-label learning,” is when a student sees an object first and then is provided with the label. This means seeing a color before being told its name. Or hearing a description of a physical force before being hearing its formal title.
The second learning procedure is “label-object learning,” the reverse order in which a student sees a label first.
Subjects in the study were asked to match the pictures of the characters with their made-up names. The presentation of information was intentionally misleading to see if learners have an easier time dealing with the inconsistency depending on the way the input was presented – either object first or label first.
The results of the study indicate that students who see objects first and then hear the name – object-label learners – process inconsistent information better than learners who hear the name first and then see the object.
Researchers detected that learners that interact with the object before hearing the name perform “frequency boosting” – the ability to process noisy, inconsistent information to identify and use the most frequent rule.
For example, when teachers interchangeably use “soda” or “pop” to describe the name of a carbonated beverage, the children who use frequency boosting will learn to use the term that is used most frequently.
A key feature of frequency boosting is that learners will also use the rule more consistently than the instructor.
“When trying to teach a child about colors, such as blue or red, not many people think about the best way to do it. People just say this is blue and point to an object. From this research, we can say that the order of presentation actually matters and that seeing the object first creates a stronger association to the name,” said Ma who conducted the research while a PhD candidate at the University of California, Irvine.
The research team also used mathematical modeling to confirm the observations as well as provide a theoretical explanation as to why one type of learner is different from the other.
“This research combines experiments with a novel mathematical model to show that object-label learners deal better with inconsistencies. It’s exciting to see that the math theory explains the observational data,” said Ma.
According to the research team, understanding how people learn could have broad applications. For example, foreign language learning programs could benefit from showing images before introducing the name of an object. The results can also be applied to math, science or any other subjects where students need to make similar associations.
Learning in natural environments is often characterized by a degree of inconsistency from an input. These inconsistencies occur, for example, when learning from more than one source, or when the presence of environmental noise distorts incoming information; as a result, the task faced by the learner becomes ambiguous. In this study, we investigate how learners handle such situations. We focus on the setting where a learner receives and processes a sequence of utterances to master associations between objects and their labels, where the source is inconsistent by design: It uses both “correct” and “incorrect” object‐label pairings. We hypothesize that depending on the order of presentation, the result of the learning may be different. To this end, we consider two types of symbolic learning procedures: the Object‐Label (OL) and the Label‐Object (LO) process. In the OL process, the learner is first exposed to the object, and then the label. In the LO process, this order is reversed. We perform experiments with human subjects, and also construct a computational model that is based on a nonlinear stochastic reinforcement learning algorithm. It is observed experimentally that OL learners are generally better at processing inconsistent input compared to LO learners. We show that the patterns observed in the learning experiments can be reproduced in the simulations if the model includes (a) an ability to regularize the input (and also to do the opposite, i.e., undermatch) and (b) an ability to take account of implicit negative evidence (i.e., interactions among different objects/labels). The model suggests that while both types of learners utilize implicit negative evidence in a similar way, there is a difference in regularization patterns: OL learners regularize the input, whereas LO learners undermatch. As a result, OL learners are able to form a more consistent system of image‐utterance associations, despite the ambiguous learning task.