Generalization and Maintenance

Why skills don't transfer, programming for generalization, and maintenance schedules

For paraprofessionals teaching skills that need to last and travel

Why this brief

A skill that works only in one room with one adult is barely a skill. A student who can put on their coat with you in the cubby corner but not at the bus line in the chaos of dismissal hasn't really learned coat-putting-on; they've learned a specific routine in a specific context. The same goes for handing in homework, asking for help, riding the bus, doing math problems, and almost every skill schools teach. Mastery in one setting often doesn't transfer; mastered skills sometimes vanish over breaks. This is generalization and maintenance — and most teaching programs underestimate how much specific work both require.

This brief covers the practical version: what generalization is, why skills often don't transfer, how to plan for generalization from the start, what maintenance schedules look like, and what to do when skills are lost. Brief 04.07 (Promoting Independence) and 04.02 (Prompting Hierarchies) cover related foundations; 06.05 (IEP Progress Monitoring) covers tracking. This brief focuses specifically on the post-mastery work that makes skills durable.

| |

| :-: |

| The frameTeaching a skill in one context isn't enough. Generalization (across settings, materials, people, situations) and maintenance (across time) require deliberate planning. They're not afterthoughts. The skill exists for the student's life, not for the teaching session. |

Who this brief is for

Paras teaching specific skills using prompting hierarchies

Paras working in self-contained, ABA-influenced, or DTT settings

Paras supporting transition-age students whose skills need to work in adult life

Inclusion paras working on skills that need to work across settings

Supervising teachers and BCBAs designing programs

What generalization means

Definition

Generalization is the ability to perform a learned skill across different contexts than where it was originally taught. Multiple kinds:

| Type | Description |

| :-: | :-: |

| Setting generalization | Skill works in different places (classroom, hallway, lunchroom, home) |

| Stimulus generalization | Skill works with different materials (different pencils, different worksheets, different books) |

| Person generalization | Skill works with different adults or peers (you, gen-ed teacher, parents, peers) |

| Verbal/cue generalization | Skill works with different prompts ("line up" vs. "time to go" vs. "let's head out") |

| Time generalization | Skill works at different times of day, days of week |

| Response generalization | Student does a similar but different version of the skill (greeting with "hi" vs. "good morning") |

Why generalization matters

Skills that work only with you don't help the student long-term

Real life requires skills that transfer

Specific contexts shift; the skill has to outlive them

The team is investing time in teaching; if it doesn't generalize, the investment is wasted

Why generalization fails

Skill was taught in only one context; brain learned the context, not the skill

Reinforcement only happened in one context

Specific cues only in one context

Discrimination not yet established

Skill is conditional and student hasn't worked out the conditions

Programming for generalization

Generalization can't be left to chance. Programs that include specific generalization planning produce skills that travel; programs that don't produce skills locked to the teaching context.

Generalization planning from the start

Don't wait until "mastery" to start working generalization

Build it into the program design

Brief 04.04 (Programming Sheets and Procedural Fidelity, planned) covers program design

Strategies for setting generalization

Teach across multiple settings — even if 80% of teaching is in one room, 20% in others

Move locations within the same room (different tables, different corners)

Eventually move to natural setting (where skill will be used in real life)

Strategies for stimulus generalization

Use multiple exemplars during teaching (different toothbrushes, different paper, different pencils)

Vary materials systematically

Don't let one specific material become the cue for the skill

Strategies for person generalization

Multiple staff teaching the same skill

Build in family training

Build in peer involvement when appropriate

Brief and train others on the protocol

Strategies for verbal cue generalization

Use multiple equivalent cues during teaching

Teach with the cue that's natural in the real setting

Vary specific phrasing

Strategies for time generalization

Practice at different times of day

Practice on different days

Don't always practice in the same time slot

Strategies for response generalization

Reinforce variations on the target behavior, not just one specific topography

Multiple acceptable responses where appropriate

"Hi," "hello," "hey," all greet appropriately

Train loosely

In some teaching paradigms, intentional variation is built into the protocol

Loose training — varying conditions during teaching — produces more durable skills

Trade-off: more variability may slow initial acquisition

Generalization probes

How do you know if generalization is happening? You probe.

What probes are

Periodic checks of skill in untrained conditions

Different setting, different materials, different person, different cue

Without prompts (or with the prompt levels expected in the natural setting)

Brief 06.01 (Data Types Overview) covers probe data

Probe schedule

Weekly probes during active teaching

Monthly during maintenance

Different probe each time — vary the dimension you're testing

Reading probe data

Skill works in untrained condition → generalization happening

Skill fails in untrained condition → generalization not happening; need explicit programming for that dimension

Variable across conditions → mixed; investigate which dimensions transfer and which don't

Common findings

Skill works with you but not with the gen-ed teacher → person generalization needed

Skill works with worksheet A but not worksheet B → stimulus generalization needed

Skill works in your classroom but not at lunch → setting generalization needed

Each finding suggests specific programming

Specific programming techniques

Multiple exemplar training

Use many examples of the same conceptual skill

"Brushing teeth" with different brushes, different paste, different sinks

"Reading a stop sign" with different stop signs in different contexts

Build the conceptual skill, not the specific instance

Train sufficient exemplars

Don't train just one or two examples

Aim for at least 3-5 examples before testing generalization

More if the dimension being generalized is complex

Train in the natural environment

Sometimes called "natural environment training" or "in vivo"

Teach the skill where it will be used

Brief 04.02 (Prompting Hierarchies) and 04.07 (Promoting Independence) overlap

Use natural reinforcers

Reinforcers that occur in the natural environment maintain the skill

"Asking for help" → getting help (natural reinforcer)

Vs. "asking for help" → token (contrived)

Brief 04.05 (Reinforcement Basics) covers natural vs. contrived

Sequential modification

If generalization isn't happening to a specific setting, train in that specific setting

Add the failed setting to the training set

Re-test

Common pattern in real programs

Mediated generalization

Teach the student strategies they can use to generalize themselves

Self-instruction, self-monitoring

"When I don't know what to do, I can ask for help"

Builds skill that transfers across novel situations

Train to generalize

Reinforce the student for generalizing

"You used the strategy here too — great"

Build the generalization itself as a behavior

Fading and generalization

Brief 04.03 (Prompt Fading) covers fading. Specific to generalization:

Fading reinforcement

Fade contrived reinforcers as natural ones take over

Move to intermittent schedules to support durability

Fading the para

Brief 04.07 (Promoting Independence)

Reduce direct support so the student does the skill without you

This is itself a generalization to "with no one prompting"

Fading specific cues

Move from explicit prompts to natural environmental cues

Teacher saying "line up" → bell ringing (natural environmental cue) → student self-monitoring

Common error

Stay in teaching mode forever — student never gets the chance to do it without support

Skill becomes locked to the teaching context

What maintenance means

Definition

Maintenance is the persistence of a skill over time after teaching has ended. A skill mastered in May should still be present in October if maintenance has happened.

Why skills are lost

Lack of practice

Reinforcement that worked during teaching no longer present

Skill not generalized to natural settings, so doesn't get used

Time, attention shifted to new skills

Specific events (illness, family stress) that disrupt

Why maintenance is overlooked

Teams move on to next skill once one is mastered

Maintenance work is unglamorous

Quick programs prioritize acquisition over durability

Until the skill is lost, no one notices

Programming for maintenance

Plan from the start

Like generalization, maintenance shouldn't be an afterthought

Build maintenance into program design

Schedule that survives the year

Maintenance schedule

After mastery, practice with thinning frequency

Daily → 3x/week → weekly → biweekly → monthly

Adjust based on data — if skill drops, increase frequency

Brief booster sessions

Short practice — 5 minutes — to refresh the skill

Less intensive than original teaching

Maintain the skill without consuming much time

Embed in natural routines

Skill that's part of natural routines maintains automatically

Brushing teeth happens daily; maintenance practice is built in

Skills that aren't naturally embedded need more deliberate practice

Track maintenance probes

Periodic checks — is the skill still there?

Probe after long breaks (winter, spring, summer)

Brief 06.01 (Data Types Overview)

Detect skill loss early

Don't wait until the skill is completely gone

Re-teach briefly when slipping starts

Earlier intervention is faster than full re-teaching

Generalization and maintenance by skill type

Self-help skills

Often need extensive across-setting and across-person generalization

Maintenance often happens through embedded daily practice

Brief 09.01 (Toileting) and other 09 series

Communication / AAC

Generalize across partners, settings, topics

Maintenance through daily use

Brief 10 series; 10.07 (Modeling AAC) covers some

Academic skills

Generalize across formats, materials, problem types

Maintenance through cumulative practice

Brief 04.12 (Reading), 04.13 (Math)

Behavioral / replacement skills

Generalize across antecedent conditions, settings, partners

Maintenance through continued reinforcement

Brief 05.06 (Functional Communication Training)

Social skills

Hardest to generalize — natural social contexts vary enormously

Maintenance through embedded daily opportunities

Brief 11.05 (Unstructured Time)

Vocational / functional

Often need community-based teaching

Generalize from school setting to community setting

Brief 11.08 (Transition 18-22)

Self-regulation skills

Generalize across antecedent conditions

Maintenance through ongoing practice and recognition

Brief 05.21 (Emotional Regulation)

Family and peer involvement

Skills that work at school but not at home, or with adults but not peers, haven't fully generalized.

Family training

Train family on the protocol when feasible

BCBA, SLP, or supervising teacher usually leads family training

Para can support carryover if family is engaged

Brief 12.09 (Working with Families)

Common challenges

Family can't always replicate the school environment

Different routines at home

Family priorities and bandwidth vary

Cultural differences in approach

Peer-mediated approaches

Train peers to be communication partners

Build skill in interactions with peers

Brief 04.18 (Peer-Mediated Strategies, planned)

Sometimes generalization to home isn't appropriate

Skills tied to school-specific contexts

Don't try to force every skill into home life

Brief 12.09 covers family relationship considerations

After breaks — common patterns

Long breaks (winter, spring, summer) often produce skill loss. Plan for it.

Common patterns

Mastered skills regress

Independence drops; prompt level needed increases

Sometimes substantial loss after long summer break

Sometimes minimal loss

Variable by student and skill

Predicting loss

Skills used during break (toileting, dressing) usually maintain

Skills tied to school routine more vulnerable

Specific academic skills often need refresh

Re-teaching after breaks

Don't restart full teaching protocol

Brief booster sessions

Run probes to assess what's there

Higher prompt levels temporarily, fading back to where you ended

Most students recover within a couple weeks

If significant loss

Skill that was solid before break and gone after may need substantial re-teaching

Treat as data — what conditions support retention?

Bring to team if pattern is consistent across breaks

Summer planning

Some students have ESY (Extended School Year) services to prevent loss

Brief 02.01 (IDEA Overview) — ESY is for students at risk of substantial regression

Family practice over summer when possible

Data for generalization and maintenance

Generalization data

Probe data — periodic checks across untrained dimensions

Setting probes — does the skill work in different rooms?

Person probes — does the skill work with different staff?

Material probes — does the skill work with different items?

Often quick — 1-3 trials per probe

Maintenance data

Periodic checks of mastered skills

Less frequent than active teaching data

Triggered by time (monthly) or by events (after breaks)

Track if skill is still at mastery level

Trends to watch

Maintenance probe shows skill still mastered → continue maintenance schedule

Maintenance probe shows skill slipping → intervene with booster

Generalization probe fails → expand training to that dimension

Documentation

Note dates of probes

Note conditions of probes

Track in the same data system as active teaching

Brief 06.05 (IEP Progress Monitoring) ties to broader reporting

Teaching the student to self-monitor

One of the most powerful tools for generalization and maintenance is teaching the student to monitor their own behavior.

What self-monitoring looks like

Student tracks their own behavior or skill use

Self-rates at end of period

Reminds themselves of the strategy

Catches their own slips

How to teach

Start with adult prompting the self-monitoring

Co-track — both student and adult record

Compare records — calibration

Fade adult role; student takes over

Why it builds generalization

Student's self-monitoring travels with them

Independent of any specific adult

Builds metacognitive skill

Brief 04.07 (Promoting Independence) overlaps

Where it works

Older students with cognitive capacity for self-monitoring

Specific concrete behaviors

Skills already partially mastered

Where it's harder

Younger students

Students with significant cognitive disability

Skills that aren't yet stable

Building generalization into the day

Naturally occurring opportunities

Don't only teach in dedicated sessions

Look for natural moments throughout the day

Embed practice in routines

Cross-staff coordination

Brief gen-ed teachers, related-service providers, specials staff

Each setting becomes a generalization opportunity

Brief 12.01 (Working with Supervising Teacher), 12.02 (Working with the Gen-Ed Teacher)

Family carryover

Skills practiced at home build generalization to home setting

Brief 12.09 (Working with Families)

Across activities

Skills practiced during academic, lunch, recess, transitions, specials

Each context provides different conditions

Brief 11.05 (Unstructured Time) and others

Pitfalls

| Try this | Watch out for |

| :-: | :-: |

| Plan generalization from the start of teaching | Treat generalization as something that happens after mastery |

| Use multiple exemplars during teaching | Train with one example and expect transfer |

| Probe across untrained dimensions periodically | Assume generalization without testing |

| Teach across settings, materials, people, and cues | Keep teaching in the same room with the same adult |

| Schedule maintenance probes after mastery | Move on to new skills and never check back |

| Plan for skill loss after breaks; refresh with brief boosters | Be surprised when students return needing reteach after summer |

| Use natural reinforcers when possible | Stay on contrived reinforcers indefinitely |

| Coordinate with family, gen-ed staff, peers for cross-setting practice | Operate as the only person teaching the skill |

| Teach self-monitoring when developmentally appropriate | Keep external monitoring forever |

| Treat skill failure in untrained conditions as data — what to teach next | Treat it as student failure |

Scenarios

Scenario 1: Skill works with you but not with the gen-ed teacher

Your student has mastered raising her hand to ask for help in your sessions. The gen-ed teacher reports she still calls out in their class.

Person generalization didn't happen. Coordinate with the gen-ed teacher to teach explicitly in their class. Specific strategies: have the gen-ed teacher cue and reinforce hand-raising in their setting; you might co-teach a few sessions to bridge; reinforce her cross-setting use when you observe it. Brief 12.02 (Working with the Gen-Ed Teacher) overlaps. Add the gen-ed setting to the formal teaching set.

Scenario 2: Skill mastered, then lost after winter break

Your student mastered tooth-brushing in November. After winter break, she needs gestural prompts for steps that were independent.

Common after long breaks. Run brief booster sessions over the next 1-2 weeks. Don't restart full teaching. Track if skill returns to baseline; usually does within a couple weeks. If it doesn't, more substantive intervention. Document for the team. Some students benefit from family practice over breaks; coordinate with family if appropriate. Brief 12.09 (Working with Families).

Scenario 3: Skill works in classroom but not at lunch

Your student has mastered asking for breaks during academic work. At lunch, when overwhelmed, she still drops to the floor.

Setting generalization needed. Practice asking-for-break in the lunchroom — even a few sessions of explicit teaching there. Brief lunch staff so they can reinforce when she does ask. Pre-teach: "At lunch, when it's loud, you can ask for a break." Sometimes the cafeteria has different antecedent conditions that need addressing — sensory, social — alongside generalization. Brief 11.05 (Unstructured Time) and 05.04 (Antecedent Strategies).

Scenario 4: Family reports skill not present at home

Your student handles morning routine independently at school. Family says he can't do any of it at home.

Different setting, different routine, different cues. Coordinate with family. Brief them on the specific protocol — what's the cue, what's the sequence, what's the reinforcer. Visit home if appropriate or supported. Sometimes the home setting is different enough that explicit programming is needed. Don't assume the family is doing something wrong; the issue is generalization, not failure. Brief 12.09 (Working with Families).

Scenario 5: Maintenance probe shows skill slipping

You ran a maintenance probe on a skill mastered three months ago. Student needed prompts that weren't needed at mastery.

Catch this early. Brief booster sessions to refresh. Increase frequency of practice for a couple weeks. Re-test in a month. If slipping continues, deeper intervention. Bring to team if persistent. Brief 06.05 (IEP Progress Monitoring) overlaps.

Scenario 6: Student doesn't generalize to peer interaction

Your student has mastered greeting adults. With peers, she doesn't initiate.

Peer generalization is often the hardest. Engineer peer interactions explicitly — pair her with a friendly peer; prompt the greeting; reinforce. Peer-mediated approaches often help. Brief 04.18 (Peer-Mediated Strategies, planned). Some students develop peer interaction more slowly than adult interaction; sustained programming over time builds it.

Closing thought

A skill that works only in one place at one time isn't really a skill — it's a routine in a specific context. Real teaching produces generalized, maintained skills that work in the student's actual life. That requires deliberate planning from the start, not as an afterthought after mastery. Multiple exemplars, multiple settings, multiple staff, fading to natural reinforcers, periodic probes, maintenance schedules, family and peer involvement — all part of producing skills that last and travel.

As a para, you're often the closest source of data on whether generalization is happening. The student you've taught a skill with is the same student who shows you whether the skill works in the gen-ed classroom, with the substitute, after the holiday, with their family. Bring that data to the team. Push for explicit programming when generalization isn't happening. The student's skill is for their life, not for your teaching session.

| |

| :-: |

| Bottom linePlan generalization and maintenance from the start. Use multiple exemplars. Probe across untrained dimensions. Teach across settings, people, materials, cues. Schedule maintenance probes. Refresh after breaks. Use natural reinforcers when possible. Coordinate across staff and family. Teach self-monitoring when appropriate. Treat untrained-condition failures as data. |

Related briefs

04.02 Prompting Hierarchies

04.03 Prompt Fading

04.05 Reinforcement Basics

04.07 Promoting Independence

04.18 Peer-Mediated Strategies (planned)

05.06 Functional Communication Training

06.01 Data Types Overview

06.03 Prompt-Level Data

06.05 IEP Progress Monitoring

11.05 Unstructured Time

12.01 Working with the Supervising Teacher

12.02 Working with the Gen-Ed Teacher

12.06 Working with the BCBA

12.09 Working with Families

Page of