Generalization vs Maintenance: How RBTs Program and Probe Both

New conditions versus later in time

Generalization is correct responding in new conditions. Maintenance is correct responding after teaching thins or time passes. For example, a child who learned to say “hi” to one technician and now greets a new teacher at school is showing generalization; still greeting that technician a month after the praise was faded is maintenance. Neither is “the skill looks pretty in the clinic.” Practice telling generalization from maintenance in the Passy RBT prep app → by naming whether the probe changed the setting or the schedule. Discrimination training builds tight stimulus control first. Generalization loosens it on purpose. Acquisition lists both tasks.

Stimulus generalization: the response occurs to similar cues (other cups, other greetings).

Response generalization: a related response appears (other ways to request).

Setting/people generalization: a new room or technician.

Maintenance: the skill still occurs next week with a thinner schedule or fewer prompts.

How the plan usually programs them

Teach with multiple examples if written. Fade prompts. Thin reinforcement. Probe in the target setting. Record the probe as a probe, not as a teaching trial, if the sheet says so.

Original teaching sketch: A mand taught only with one bubble container may not occur with a different bottle at grandma’s house. That is a generalization miss, not “noncompliance.” A mand that occurs in the clinic every session but disappears after two weeks of thinner tokens is a maintenance miss.

DTT versus naturalistic teaching often matters because naturalistic teaching can program generalization earlier. That is a plan choice, not a technician improvisation.

Prompting that never fades will look like a clinic-only skill.

When generalization and maintenance still swap, try probe questions in the Passy RBT prep app that change only one condition →.

When maintenance probes are conducted

A maintenance probe checks whether a mastered skill still occurs without the teaching that built it. Most plans start maintenance probes after the target meets the mastery criterion written in the program, then repeat them on the schedule the supervisor sets. The RBT outline does not set a probe frequency; the plan does. Many programs space probes further apart while the skill holds, for example weekly at first, then every two weeks, then monthly.

Run the probe the way the sheet defines it, which usually means no prompts and no teaching trials, and record it as a probe. If the client misses a probe, report it. The supervisor may move the target back into teaching or change the schedule. Do not reteach on the spot unless the plan says to.

Generalization probes follow the same logic but change the condition instead of the time: a new person, setting, or set of materials.

Role limits

Do not take the client to an unplanned community setting because you personally want to “generalize.” Follow the plan and supervision.

Do not secretly keep a continuous reinforcement schedule after the plan thinned it, just to keep the graph pretty. That delays maintenance.

The what-to-study guide keeps these inside Domain C.

If probe types are the miss, review missed generalization and maintenance questions in the Passy RBT prep app →.

A perfect table skill that never leaves the table is unfinished teaching.

Then mix generalization and maintenance into Domain C practice in the Passy RBT prep app → beside discrimination items so you know when tightness is the goal.