Goals Of Behavior Analysis As A Science

The Actual Work of Behavior Analysis

Most people who come into behavior analysis expecting it to be some sort of transformational technique that fixes human problems in a weekend. It isn't. It's a scientific discipline with a fairly narrow set of goals, and honestly, most practitioners spend the bulk of their time just trying to describe things accurately enough to measure them. The Goals Of Behavior Analysis As A Science have been pretty stable since Sid Skinner laid out the framework in the sixties. There are really four of them, and they build on each other in a specific order. You can't skip ahead. People always try to skip ahead.

Description Comes First and It Is The Hardest Part

The first goal is to describe behavior precisely. Not loosely. Not "the kid was aggressive" or "the employee wasn't motivated." Those are interpretations dressed up as observations. Description means you define the topography of the behavior so clearly that two different people watching the same event would count it the same way. Same frequency. Same duration. Same intensity measures. I worked with a clinic for about three years where they were running an ABA program for autistic children and couldn't figure out why their data looked so inconsistent across therapists. It turned out their definition of "manding" was different depending on who was collecting data. One therapist counted any vocalization toward a desired item. Another only counted clear word approximations. They were collecting completely different behaviors without realizing it. The fix was rewriting the operational definition with concrete examples and counterexamples, then running reliability checks weekly. Data stability went from about sixty percent inter-observer agreement to over ninety in six weeks. This is where most programs fail before they even start. If you can't describe what you're measuring, nothing else matters. Prediction requires it. Explanation requires it. Influence requires it.

Prediction Is Where The Rubber Meets The Road

Once behavior is described consistently, the second goal is to predict when and where it will occur. This isn't about guessing. It's about identifying functional relationships between environmental variables and the behavior. You're looking for patterns. Antecedents. Consequences. Setting events. The variables that reliably precede or follow the occurrence. The thing beginners miss is that prediction doesn't require understanding causation. You can have a solid predictive model and still be completely wrong about why something happens. That's fine. That's how science works. In practice this means you might find that a particular student consistently engages in off-task behavior fifteen minutes into math instruction, regardless of the math content or the teacher. The prediction is solid. The explanation might involve task difficulty, attention-seeking, escape motivation, or fatigue. You don't need to know which one yet to use the prediction. My own experience with this came when I was consulting on a behavioral program at a residential facility. Staff kept attributing a resident's self-injurious behavior to "frustration" because that was the story that made sense narratively. But the data showed the behavior occurred almost exclusively after transitions between activities, not during challenging tasks. The narrative explanation was wrong. The predictive model was right. Correcting the narrative didn't change the intervention, but it changed how staff responded during transitions, and that made a real difference.

Explanation And Why It Is Overrated

The third goal is explanation. This is the one everyone thinks matters most, and in practice it often matters the least for day-to-day work. Explanation means identifying the functional relations that produce the behavior. What variables control it. What maintains it. In behavior analysis this usually means a functional analysis or at minimum a functional assessment. Here's the counterintuitive part: you can often produce meaningful change without a complete explanation. I've seen interventions succeed based on solid predictive data alone. The behavior decreased because the antecedent conditions were modified, even though the exact motivational function wasn't fully specified. Conversely, I've seen perfectly executed functional analyses that led to interventions which failed because the explanation was right but the implementation was sloppy. A good explanation is valuable, but it's not a guarantee of success. The downside of over-investing in explanation is that it can delay action. Functional analyses take time. They require trained personnel. They sometimes produce risky behavior during the assessment phase. If you're working in a school setting with limited resources and a child who's been sitting on a waitlist for six months, spending eight weeks on a comprehensive functional analysis before doing anything might not be the most ethical use of your time. Partial assessments, brief trials, and iterative adjustments often get better real-world results.

Influence Is The Final Goal But It Requires Restraint

The fourth goal is to influence behavior. This is where applied behavior analysis lives, and it's also where the field gets the most criticism. Influence doesn't mean control in the authoritarian sense. It means systematic intervention based on the descriptive and predictive data you've already collected. The practical challenge here is that influence requires changing environment, not just the individual. Most people want interventions that target the person with the behavior. The science says that approach is usually insufficient. You modify antecedents. You change consequence structures. You alter the physical and social environment. This is harder to sell to administrators and parents who want something done to the person rather than to the context. I ran into this directly when working with a school district that wanted a behavioral specialist to "fix" a student who was disrupting classes. The data showed the behavior was maintained by peer attention and occurred most frequently during unstructured times. The simplest intervention would have been to adjust scheduling and teach alternative social skills. Instead, the district wanted a one-on-one aide and a behavior plan focused entirely on the student. We spent six months fighting for the environmental changes. The student's behavior improved by about forty percent after the schedule adjustment alone. The aide and behavior plan added maybe five percent more improvement on top of that.

There are scenarios where influence through environmental modification simply won't work fast enough. Acute safety situations sometimes require more direct individual-focused interventions as a temporary measure. The field acknowledges this. It's not a failure of the science, it's a recognition that real-world constraints exist. But the default should always be environmental first, individual second, and the justification for reversing that order should be explicit and time-limited.

What The Field Gets Wrong About Its Own Goals

Behavior analysis as a science has a clean theoretical structure. The four goals map logically onto each other. Description, prediction, explanation, influence. In practice the mapping gets messy. People chase influence before they've secured description. They treat explanation as the end goal rather than a means to better influence. They confuse correlation with functional relation all the time. The biggest ongoing problem is that the field still struggles with generalization. A behavior change that works in a clinical setting often disappears when it moves to a natural environment. Not always, but frequently enough to be annoying. The goals are designed to produce durable change, but durability depends heavily on whether the maintaining variables have actually been addressed or just suppressed temporarily. If you reduce a behavior by removing its reinforcement but the person still encounters the same motivating operations in the wild, the behavior will likely return. Another nuance that doesn't get enough attention is that these four goals apply differently depending on whether you're doing basic research or applied work. Basic behavior analysts might spend decades on description and explanation without ever pursuing influence. Applied analysts are expected to hit all four but often have to compress the timeline. Neither approach is wrong. They're just different applications of the same scientific framework.

The Goals Of Behavior Analysis As A Science aren't aspirational. They're operational. If you're not describing, predicting, explaining, and influencing systematically, you're not doing behavior analysis. You're doing something else that might resemble it, but the resemblance is what gets people in trouble.