Last changed 17 Mar 2008 ............... Length about 2,000 words (17,000 bytes).
(This document started on 15 Mar 2008.) This is a WWW document maintained by Steve Draper, installed at You may copy it. How to refer to it.

Web site logical path: [] [~steve] [this page]

Christine Howe and constructive interaction

Contents (click to jump to a section)

by Steve Draper

Key Points established by this field of research

This is a short briefing on Howe's work on constructive interaction in children. "Constructive interaction" is not a term she uses, but is Miyake's term for the way when two people discuss a concept they may both improve their understanding of it. It is, it seems to me, a fundamental and in some ways counterintuitive feature of human life and learning; and one that should be at the heart of university life, and fundamental to well organised learning environments. The most detailed experimental work I know of has been by Howe and her collaborators.

The key points (to me) are:

  1. The pre-conditions that must be set up to get conceptually productive peer discussion (because it often isn't, but it can be made, reproducibly productive).
    1. Different conceptions, in both level and detail [best ref: 1990 ]
    2. Different predictions [best ref: 1995a ]
    3. Public? commitment to a specific prediction/view at the start. [best ref: 1990 ] "publicly associated with a particular viewpoint". The learners have to individually predict in writing; then have to publicly display these and agree.
    4. Teachers shouldn't direct the discussion either by saying what to pay attention to, or by pushing them to your conclusion. [best ref: 1993 ]

  2. The learning gains show only at the delayed post-test [best ref: 1990, 1992a, 2005 ]
    And this works by priming them to get more out of later inputs [2005]
    I.e. the learning does NOT usually happen during the discussion. So the discussion must implant a delayed cause; and the learning happens later (but does NOT happen without the prior discussion).

  3. 1992 "not agreement but private conflict resolution" 2000a Advance from/when conflicts resolved (in dlog/group)
    I.e. this is not about consenus, nor joint idea building, nor a common product. It is mutual help with two private tasks, private building of understandings. This is actually like most conversations, where the benefits are typically rather different between the participants. It is like eating together at a restaurant; it is not like cooperating to build a house, or sharing a car to make the same journey.

  4. Shows that the more advanced child also advances; [best ref: 1990 ]
    and that the 2 peers get different things out of the same interaction;
    That social agreement doesn't tell you they privately agree; and that more advanced kids regress in the groups (but rebound further by post-test).
    and so too the importance methodologically of doing both group (transcript) and solo interviews as measures. [best ref: ?? ]
    I.e. learning is NOT by transmission or social consensus (contrary to what you might imagine from Vygotsky, Lave, social constructivism, etc.).

Method (typically)

[best ref: 1990 ]
By method I mean: how do you set the conditions for peer discussion to be productive of learning (because it often isn't, but it can be made reproducibly productive).

Solo prediction then group (to get public commitment)
Task: predict, run test, explain actual result.
For groups: agree the prediction (then test) agree the explanation.
Multiple cases i.e. varying variables within a fixed paradigm of test, each case dictated via a booklet by experimenters.

Group size 4.

Writing down (OE) conclusions at end of group work is important; and selecting between ones from researcher is bad. 1993, 2003


Williams & Tolmie (2000a), at least, contrast feedback and peer discussion as causes. Their experiment shows peer discussion is more powerful for learning (conceptual progress) than feedback. Partly this is consistent with the full feedback loop requiring more constructive help, not just right/wrong answer. But partly it is that the conceptually helpful task isn't just doing a test or something, but generating free-wheeling conclusions/hypotheses.


The value of groups, at least for learning, is not about group work substituting for individual ("solo") work; but rather it is about the interaction of the two e.g. with questions activiated in discussion leading to different private work later.

Prior solo task to get public commitment: 1990??
Diffs between social agreement and private interviews
The way most benefits there only after delayed post-test i.e. group itself doesn't do it alone, but is catalyst for future solo development.

N.B. the task instruction is usually for the group to agree (as if it were an action group).

N.B. EVS can directly implement this solo commitment phase, preceding public discussion.

2000, 2003 take solo:group combination forward; and relate it to concept vs. procedural learning.

Writing down group decisions/consensus is also very important for learning (returning to solo mode??).

And 2003 emphasises how small differences in procedure make big differences: so it is not solo vs. group in general, but the details of what is done in groups.

Progress does not depend on agreement. 2000a, 1990
The "solutions" with most agreement in a group were the most conceptually advanced.
But the further advance at delayed post-test shows it is not a pre-condition for advance; and false agreement i.e. regression could then lead to further advance. So a lot of advance has NO dependence on agreement.
(It may be that measures of agreements in groups are measures of fuller engagement with the persuasion/explanation task, and it is this that is productive of learning.)

Annotated bibliography: Howe References

This is in chronological, NOT alphabetical, order

Miyake,N. (1986) "Constructive interaction and the iterative process of understanding" Cognitive Science vol.10 no.2 pp.151-177

Howe, C.J., Rogers,C. & Tolmie, A, (1990) "Physics in the primary school: Peer interaction and the understanding of floating and sinking" European J. of psychology of education vol.V=5 no.4 pp.459-475
[This can be hard to get hold of]
[Established method of individual then solo i.e. prior public commitment, to get productive discussion.
Delayed post-test, 5 weeks, solo, in study 1; in study 2 2-week delay. No rel. of change to pre-test i.e. advanced also advanced further. Compared 3 groups: similar views, differing in both level and view within a level, differing only in view but same level. N=121 Mean increases at best 10% of the scale. Already it shows that advanced kids regress in group, then overtake their pretest by delayed post test. And that progress correlates with nmb. of agreements within the group.
Group size 4; 7+7+7 groups. Pool = 121, participants = 84?]

Howe, C.J., Tolmie, A., & Anderson,A. (1991a) "Information technology and group work in physics" Journal of Computer Assisted Learning vol.7 no.2 pp.133-143
[Joint decision making is the prompt for change.]
12-15 years old; computer presented, but group work. Falling objects (and wind resistance). Predict and explain. Solo paper test; 5 week gap; pairs at computer with their paper work, simulation/animation, and computer presented tasks; 1 week gap; paper post-test.
No gender diff in overall effects, but in method of discussion. Males: they notice difference in prediction, then discuss preconceptions, then advance their conceptions. Female: no rel. between predict and expl; but discussing expls caused advance. MF pairs discussed less, and resolved less.

Howe, C. J. (1991b) "Explanatory concepts in physics: towards a principled evaluation of teaching materials" Computers and Education vol.17 no.1 pp.73-80
[Socratic dlog; and methodology. Arguing in favour of ctrl. expts. and theory.]

Howe, C.J., Tolmie, A, and Rogers,C. (1992a) "The acquisition of conceptual knowledge in science by primary school children: Group interacting and the understanding of motion down an incline" British Journal of Developmental Psychology vol.10 pp.113-130
[Not agreement but private conflict resolution. Diff. start conceptions. 4-week delayed post-tests. A source of their std. method; particular stress on non-social-agreement and delayed digestion]

Howe, C.J., Tolmie, A., Anderson,A. & MacKenzie,M. (1992b) "Conceptual knowledge in physics: The role of group interaction in computer-supported teaching" Learning and Instruction vol.2 pp.161-183
[HE not primary school. Peer conflict promotes learning. 3-week delayed post-test. Group and solo scores.]

Tolmie, A., Howe, C J, Mackenzie, M. and Greer, K. (1993) "Task design as an influence on dialogue and learning: Primary school group work with object flotation" Social Development vol.2 no.3 pp.183-201.
[Pre-conditions: not too much external direction to the conclusion or of Ls' attention, but active free-wheeling generation by Ls of conclusions ]

Tolmie, A. & Howe, C.J. (1994). "Computer-directed group activity and the develpoment of children's hypothesis testing skills" In H.C. Foot, C.J. Howe, A. Anderson, A. Tolmie & D. Warden (Eds.) Group and Interactive Learning pp.139-144 (Southampton: Computational Mechanics Publications).

Howe, C.J. & Tolmie, A. (1994). "Task design: a neglected variable in the study of group work" In
H.C. Foot, C.J. Howe, A. Anderson, A. Tolmie & D. Warden (Eds.) Group and Interactive Learning. pp.429-435. (Southampton: Computational Mechanics Publications).

Howe, C.J., Tolmie, A. and Mackenzie, M. (1995a) "Computer support for the collaborative learning of physics concepts" in Computer supported collaborative learning pp. 223-243 (ed.) C.E. O'Malley (Springer Verlag: London) (from a 1989 workshop).
[4 school studies. For benefit, must differ both / not only in concepts but also in the predictions they make. Used delayed post-tests.]

Howe, C.J., Tolmie, A., Greer, K. and Mackenzie, M. (1995b) "Peer collaboration and conceptual growth in physics: task influences on children's understanding of heating and cooling" Cognition and Instruction vol.13, no.4 pp.483-503.
[Detailed study of some aspects of mechansim/process. Depends on the type of the prior conception: whether it revolves around process or not.]

Howe, C.J., Tolmie, A. & Rodgers, C. (1996) "The acquisition of conceptual knowledge in science by primary school children" In L.Smith (Ed.), Critical Readings on Piaget. (London: Routledge)

Howe, C.J. & Tolmie, A. (1998a) "Computer support for learning in collaborative contexts: prompted hypothesis testing in physics" Computers & Education, vol.30 no.3-4 pp.223-235.
[9-14 yr old. Prompts them through hyp testing. 2 refs by Wood on trouble prompting.]

Howe, C.J. & Tolmie, A. (1998b) "Productive interaction in the context of computer-supported collaborative learning in science" In K. Littleton & P. Light (Eds.), Learning with Computers: Analysing Productive Interaction pp.24-45 (London: Routledge)

Tolmie, A. & Anderson, A. (1998c) "Information technology and peer-based tutorials" The Psychologist vol.11 pp.381-384
[psych. Tutorials with no tutor, agenda run by a computer.]

Foot, H.C., & Howe, C.J. (1998d) "The psychoeducational basis of peer-assisted learning" ch.2 pp.27-43 in Topping,K. & Ehly,S. (eds.) Peer-assisted learning (LEA)
[Really about PAL. Collaborative learning vs. peer tutoring (CL vs PT). Notes in TM PAL7.]

Williams, J.M. & Tolmie, A. (2000a) "Conceptual change in biology: group interaction and the understanding of inheritance" British Journal of Developmental Psychology vol.18 no.4 pp.625-649.
[Biology. 8-12 year olds. Diff. prior concepts. Resolution of conflicts important to advance. Peer discussion better than feedback: written right answer assertions after they'd committed to a answer.]

Howe, C.J., Tolmie, A., Duchak-Tanner, V & Rattray, C. (2000b) "Hypothesis testing in science: group consensus and the acquisition of conceptual and procedural knowledge" Learning and Instruction vol.10 no.4 pp.361-391
[9-12 year olds. Domain is shadow size caused by diff. brightness, distance etc.
Issue is proc. vs. conceptual learning.
Proc requires contingent tutoring from an expert; but conceptual growth requires peers, no authorities, differing views.
Her wheeze here is to require consensus as a bridge, and get the best of both. So they do group without T first; get consensus; then T does scaffolding on a proc task. Task was deciding how to do a good expt. test.
It's also about a) metacog; b) hyp testing; c) explanation vs. prediction vs. acting.
Metacog: realising that the exercise was a test, not just a task. Her arg. is that consensus was important in getting the Ls (the group) to make that connection and keep it in mind while acting, and while getting contingent shaping from T. For her cog. engagement is doing the task; metacog engagement is relating the task to the concept, to it being a hyp-test. And it's about the kids "owning" the test and the hyps; which in typical CT interactions they don't. I.e. the issue Piaget was getting at is that adults poison fruitful debate because the Ls don't own the expls and hyps. Or to put it another way: real uncertainty about the outcome and the concept.]

Anderson, A., Howe, C., Soden, R., Halliday, J. & Low, J. (2001) "Peer interaction and the learning of critical thinking skills in further education students" Instructional Science vol.29 pp.1-32
[CT teaching in FE. Not very successful.]

Howe, C.J. & Tolmie, A. (2003) "Group work in primary school science: discussion, consensus and guidance from experts" International Journal of Educational Research vol.39 pp.51-72
[Directly about group vs. solo. Primary schools. Again, proc. vs. conceptual. Repeats 2000b with different domain. Domain is heat transfer. 2 week delayed post-test.
Some diff. results but basically the recipe still works. "The centrality of mechanisms over evidence"
Recipe p.68: a) Solo predictions b) group discuss, and reach joint decision. c) Feedback on this from an expt. d) Joint interpretations (about which are the causal factors) e) Write records of this: of which factors important, which irrelevant. T used to keep group on track with this proc.
In fact, she says this suggests there is some value for group work for proc learning (at least this kind of proc: hyp testing). ]

*Howe, C., McWilliam, D. & Cross, G. (2005) "Chance favours only the prepared mind: incubation and the delayed effects of peer collaboration" British Journal of Psychology vol.96 no.1 pp.67-93
[Given delayed effect, which of 3 cog. theories explains it? 2 don't; being now primed to respond to later new input differently does.
Primary school. Floating, sinking. Pre-test; collab task; 3 demonstrations after 2,4,6 weeks; and at 8 weeks post-test. Cmp those who did/not get the collab task. Clear effect.]

Howe, C. & McWilliam, D. (2006) "Opposition in social interaction between children: why intellectual benefits do not mean social costs" Social Development, vol.15 no.2 pp.205-231
[Start work on aggression vs. benefit of peer conceptual conflict]

Web site logical path: [] [~steve] [this page]
[Top of this page]