1. Assessing the efficacy of fluency training in relating relations for increasing performance on general cognitive ability
Shane McLoughlin
Ian Tyndall
Antonina Pereira
University of Chichester

2. Outcomes of SMART training from other studies
Remediation of low IQ children aged 9-12; 2-3 sd rise in neuro-typical children (Cassidy, Roche, & Hayes, 2011)
23 point average IQ rise among year old children (Cassidy, Roche, Colbert, Stewart, & Grey, 2016)
Improvements in spelling, reading, and numerical operations higher in SMART group than computer coding group aged (Hayes & Stewart, 2016)
Remediation of Alzheimers Disease (Presti, Torregrossa, Migliore, Cumbo, & Roche, 2016)
Increases in reaction speed on intellectual tasks (McLoughlin, Tyndall, & Pereira, 2016)
Lesson descriptions should be brief.

3. Objectives
To test the efficacy of relational responding interventions for increasing intelligence
Simple and complex relational training procedures
To investigate the role of response latency in relation to intelligence
Experimental vs control
Measured at 4 time points
To investigate fluency of relational responding in relation to intelligence
IQ metrics only
Response latencies only
Example objectives
At the end of this lesson, you will be able to:
Save files to the team Web server.
Move files to different locations on the team Web server.
Share files on the team Web server.

4. SMART Strengthening Mental Abilities with Relational Training www
SMART Strengthening Mental Abilities with Relational Training
YES / NO counterbalanced
Reinforcement provided during training block (16 trials), not provided during testing block (16 trials)
Starts with affirmation of presented relation, proceeds towards reversals
Starts with SAME / OPP symmetrical relations
Progresses to MORE / LESS asymmetrical comparative relations
WUG is the same as DAX Is WUG the same as DAX? NO YES HEP is opposite to LEK Is LEK opposite to HEP? YES NO

5. How presentation will benefit audience: Adult learners are more interested in a subject if they know how or why it is important to them.
Presenter’s level of expertise in the subject: Briefly state your credentials in this area, or explain why participants should listen to you.

6. SMART Strengthening Mental Abilities with Relational Training www
SMART Strengthening Mental Abilities with Relational Training
LER is more than FEL FEL is more than ARP TUJ is less than ARP TUJ is more than JUB Is JUB less than LER? NO YES
Increases in complexity
Participants instructed to play “Brain Flexibility” mini-game

7. SMARTA Strengthening Mental Abilities with Relational Training – for Analogical Responding
Complex relational training program on relating / differentiating stimulus relations (see McLoughlin & Stewart, in press)
Deriving relations based on arbitrary properties of stimuli
SAME / OPPOSITE networks related to SAME / OPPOSITE networks
MORE / LESS networks related to MORE / LESS networks
In all cases, the relation between relations was always either SAME or OPPOSITE

8. SMARTA Strengthening Mental Abilities with Relational Training – for Analogical Responding
Responding in accordance with relations between relations
YES / NO counterbalanced
Starts with assessment of 46 unique trial types
Participants instructed to train in trial types answered incorrectly in initial assessment
Reinforcement provided during training block (16 trials), not provided during testing block (16 trials)
After training, instructed to practice block of 46 trial types
LER is more than FEL FEL is more than ARP TUJ is less than ARP TUJ is more than JUB Is TUJ to LER opposite to FEL to JUB? NO YES

9. SMARTA Interface Menu

10. Table 1
Demographics
Participant
Gender
Age
SLD P1
Male 26 NO P2
Female 39 P3 42 P4 52 C1 28 C2 29 C3 24 C4 22
Note. SLD = Previously diagnosed Specific Learning Difficulty.

11. Table 2
Procedure
Participant
Time1
Time 2
Time 3
Time 4 P1 --
Begin SMART
End SMART
Begin SMARTA
End SMARTA P2 P3 P4 C1 C2 C3 C4
Note. SMART = Strengthening Mental Abilities with Relational Training; SMARTA = SMART for analogical responding.
Kaufman Brief Intelligence Test administered at the beginning of each session.
20 minute test.
Mean split-half reliability coefficient across all subtests = .93; K-BIT FSIQ / WAIS FSIQ correlation coefficient = .75 (Kaufman & Kaufman, 1990).
Participants proceeded at their own pace through a computerised version of the test.

12. Results A series of 2*4 mixed ANOVAs were used to analyse results.
Condition (2) (Between subjects factor)
Experimental, Control
Time (4) (Within subjects factor)
Time 1, Time 2, Time 3, Time 4
All post-hoc tests used a Bonferroni correction
Dependent Variables
Intelligence (Verbal, Non-Verbal, Full Scale)
Response Latency (Verbal Knowledge, Matrices, Riddles; Full K-BIT)
Response Fluency (Verbal Knowledge, Matrices, Riddles; Full K-BIT)
Fluency = answers correct / response latency

13. Overall Fluency changes

14. Fluency change across all subscales of the K-BIT
Observed power for experimental participants = .635
Observed power for control participants = .531
sig
non-sig
non-sig
sig
sig
non-sig

15. Fluencies (percentage of correct answers divided by response latencies)
Full K-BIT
Significant main effect of Time on fluencies: F(1.125, 6.809) = 16.03, MSE = .005, p = .002, ηp2 = .728
No significant main effect of Condition on fluencies
Significant main interaction effect of Time and Condition: F(1.125, 6.809) = 7.30, MSE = .002, p = .029, ηp2 = .549
Pairwise Comparisons
Experimental
Time 1-2 No significant differences in fluencies
Time 2-3 Significant difference in fluencies: F(3, 4) = 6.80, p = .009, ηp2 = .836
Time 3-4 Marginally significant difference in fluencies: F(3, 4) = 6.80, p = .05, ηp2 = .836
Time 2-4 Significant difference in fluencies: F(3, 4) = 6.80, p = .017, ηp2 = .836
Control
Time 1-2 Significant difference in fluencies: F(3, 4) = 5.32, p = .019, ηp2 = .800
Time 2-3 No significant differences in fluencies
Time 3-4 No significant differences in fluencies

16. Fluencies (percentage of correct answers divided by response latencies)
Verbal Knowledge
Significant main effect of Time on fluencies: F(1.23, 7.38) = .02, MSE = .006, p = .002, ηp2 = .777
No significant main effect of Condition on fluencies
Significant main interaction effect of Time and Condition: F(1.23, 7.38) = , MSE = .006, p = .012, ηp2 = .627
Pairwise Comparisons
Experimental
Time 1-2 No significant differences in fluencies
Time 2-3 Significant difference in fluencies: F(3, 4) = 16.80, p = .001, ηp2 = .93
Time 3-4 Marginally significant difference in fluencies: F(3, 4) = 16.80, p = .055, ηp2 = .93
Time 2-4 Significant difference in fluencies: F(3, 4) = 2.38, p = .006, ηp2 = .64
Control
Time 1-2 Significant difference in fluencies: F(3, 4) = 4.80, p = .03, ηp2 = .78
Time 2-3 No significant differences in fluencies
Time 3-4 No significant differences in fluencies

17. Fluencies (percentage of correct answers divided by response latencies)
Riddles
Significant main effect of Time on fluencies: F(3, 18) = 16.82, MSE < .001, p < .001, ηp2 = .737
No significant main effect of Condition on fluencies
No significant main interaction effect of Time and Condition

18. None for Verbal Reasoning, Matrices, or Riddles subscales of KBIT-2
Accuracy changes
None for Verbal Reasoning, Matrices, or Riddles subscales of KBIT-2

19. Response latency changes

20. Response Latencies (number of seconds required to completion)
Verbal Knowledge
Significant main effect of Time on RLs: F(1.292, 7.752) = , MSE = , p < .001, ηp2 = .84
No significant main effect of Condition on RLs
Significant main interaction effect of Time and Condition: F(1.292, 7.752) = , MSE = , p < .001, ηp2 = .685
Pairwise Comparisons
Experimental
Time 1-2 No significant differences in RLs
Time 2-3 Significant difference in RLs: F(3, 4) = 25.69, p < .001, ηp2 = .95
Time 3-4 Significant difference in RLs: F(3, 4) = 25.69, p = .021, ηp2 = .95
Control
Time 2-3 No significant differences in RLs
Time 3-4 No significant differences in RLs

21. Response Latencies (number of seconds required to completion)
Riddles
Significant main effect of Time on RLs: F(1.349, 8.095) = , MSE = , p = .01, ηp2 = .626
No significant main effect of Condition on RLs
No significant main interaction effect of Time and Condition

22. Response Latencies (number of seconds required to completion)
Full K-BIT
Significant main effect of Time on RLs: F(3, 18) = 24.43, MSE = , p < .001, ηp2 = .803
No significant main effect of Condition on RLs
Significant main interaction effect of Time and Condition: F(3, 18) = 10.24, MSE = , p < .001, ηp2 = .630
Pairwise Comparisons
Experimental
Time 1-2 No significant differences in RLs
Time 2-3 Significant difference in RLs: F(3, 4) = 13.42, p = .004, ηp2 = .91
Time 3-4 Significant difference in RLs: F(3, 4) = 13.42, p = .037, ηp2 = .91
Time 2-4 Significant difference in RLs: F(3, 4) = 2.38, p = .002, ηp2 = .91
Control
Time 2-3 No significant differences in RLs
Time 3-4 No significant differences in RLs

24. Limitations
No experimental control over learning opportunities for Verbal Knowledge and Riddles subscales.
Short IV did not increase performance on Matrices subscale.
No measure of relational skills fluency.
SMARTA involves mostly the same DRR tasks as SMART so levels of “relational training” IV are not easily contrasted.

25. Conclusions
Response latencies decreased more noticeably than accuracy scores increased in this short study.
Fluency (correct answers / response latency) may be an important measure of intelligent behaviour that is currently not included in many traditional psychometrics.
Speed and accuracy of relational responding across multiple exemplars may be a more realistic and useful measure of intelligent behaviour – subject to further testing
SMARTA provides first analogical responding precision teaching software. Not tested: Relating temporal and hierarchical relations.

26. Assessing the efficacy of fluency training in relating relations for increasing performance on general cognitive ability
Shane McLoughlin
Ian Tyndall
Antonina Pereira
University of Chichester