Hey Alexa, Do My Homework!

Co-authored by Dr. Melissa Swisher, Lecturer, Purdue University

We rely on virtual assistants for many mundane activities–getting directions, calling a place of business, performing simple mathematical operations, and answering a multitude of questions. Amazon’s Alexa™ goes a step further and helps with homework. Does that mean students asking Amazon’s Alexa™ for help with their homework are sidestepping the point of homework:  Are they missing out on an opportunity to learn how to solve problems?

When a student asks Amazon’s Alexa™ (or Google Assistant™ or Siri™) for information, they get an answer to a specific question. For instance, a student might need to know who invented the operant chamber. When asked, “Alexa, who invented the operant chamber?” Alexa™ responds, “B. F. Skinner invented the operant chamber.” Then, the student can write Skinner’s name in the blank or use that information to solve another problem or question. The student will now know that Skinner invented the operant chamber. However, by asking Alexa, the student missed the opportunity to engage in a few steps that might be considered part of the problem-solving process. For example, before we could ask Alexa™, we would have to find another way to get to the answer. We might have had to determine which unit covered the operant chamber, looking back through the textbook or class notes about early operant conditioning to find the answer. Learning how to find an answer without asking someone or something might be just as important as getting the right answer to a question, despite basing students’ grades on simply providing the right answer. That is, we can only acquire new knowledge by asking someone (or something). But what if we want to know something that is still unknown?

To qualify as problem solving, we must 1) encounter some barrier that prevents solving the problem, 2) be able to emit a response that produces a reinforcer (i.e., a response must be in the learner’s history or their repertoire), and 3) recognize (or discriminate) that there is a problem in an environment that supports this behavior (motivation; Kieta, Cihon, & Abdel-Jalil, 2018; see also Donahoe & Palmer, 1994). Although we can solve the problem with a response in our repertoire, the solution might require a combination of responses via contingency adduction (Andronis, Layng, & Goldiamond, 1997; Layng, Twyman, & Stikeleather, 2004; Street & Johnson, 2014).  As an example, we might need to respond to a person asking how old we are. We might not remember how old we are even if, with a little time, we could get to the answer. Ingvarsson and Hollobaugh (2010) taught four boys with autism how to respond to these can’t answer questions with “I don’t know; please tell me.” Then the adult supplies the answer, and finally the child can appropriately respond when asked again. For a more complex task, a person might need to know how to connect a new television. Her previous knowledge of setting up computers and other electronics should prompt her to look for cables, input/output connections, and an electrical outlet for the television. The steps for setting up a computer versus a television won’t be the same, but the component skills might be.

Problem-solving skills in mathematics are often the target of educational interventions. Sherman and Bisanz (2009) found that many elementary school learners had difficulty understanding the meaning of the equal sign (“=”) in problems with equalities (e.g., 1 + 3 = 2 + _). Thankfully, there are many ways to establish mathematical skills with behavior analytic instruction (e.g., Hofstadter-Duke & Daly, 2015; Lynch & Cuvo, 1995; Rippy & Doughty, 2017).

Other problems may require several precurrent responses (Levingston, Neef, & Cihon, 2009; Polson & Parsons, 1994; Skinner, 1984) before reaching a solution. For some problems, we might use a talk-aloud procedure: “To get to West Lafayette, you take I-35; no, that’s in Texas. You start at I-57, then take…the exit in Champaign…that’s I-74. You follow that until the exit around 30–maybe 34. Yes, that’s right.” This interstate-highway intraverbal of 57-74-231 might be visual bidirectional naming if the speaker-as-listener imagines driving the route and seeing the signs as she gives directions (Miguel, 2018; see also Weisenburgh-Snyder, Malmquist, Robbins, & Lipshin, 2015 and Whimbey, Lockhead, & Narode, 2013). Teachers can guide learners through overt talk-aloud problem-solving strategies (see Kieta, Cihon, & Abdel-Jalil, 2018) that proficient learners can use covertly to efficiently select appropriate responses. In fact, Axe, Phelan, and Irwin (in press) analyzed 12 empirical studies that referenced Skinner and problem solving. Various authors taught children and young adults behavior chaining, self-prompting, visual imagining, recombining units, and sorting for math, spelling, play, social, and verbal behavior skills.

There is no one-size-fits-all approach to teach problem solving. However, instruction in a general, strategic approach to solving a few specific problems can help learners get started when confronted with a new problem. When Alexa™ fails learners–and she will eventually–they can successfully recombine units of their own repertoire to solve seemingly impossible problems; even on their homework – and that’s something behavior analysts are learning how to teach!