Joel Michael
Allen Rovick
October, 1996


MacMan (Dickinson et al., 1973) is a mathematical model of the baroreceptor reflex. It was originally written as a FORTRAN program and ran on the main-frame computer at McMaster University where it was developed.

MacMan had a simple interface that allowed the user to specify the parameters to be changed in running the model, and to designate what variables would be outputted (graphically or in table form) as the model ran (all output was via a lineprinter).

We first ran MacMan over the phone (using an acoustic coupler modem). We were sufficiently impressed that we requested permission to translate the FORTRAN code into TUTOR, the language for the PLATO system. We were given permission to do this, and a print copy of MacMan was sent to us. The Office of Computer-Based Education (as it was then called) at Rush carried out the translation.

We used MacMan (on the PLATO system) for only a year or so. We wrote a lab manual that described certain experiments to be carried out, provided a description of how students could design and carry out their own experiments, and included a set of questions to direct the students' attention to certain phenomena that they should observe.

However, MacMan proved to be a not particularly useful learning resource for our students. There were several reasons for this failure. (1) Students were likely to "cook-book" the experiments (carry them out in a rote fashion without any thought about what they were doing, why they were doing it, and what the results might mean) just as they frequently did with "wet lab" experiments. (2) Our students were so inexperienced in experimental design, and knew sufficiently little about the physiology, that they were unable to generate experiments that would aid them in understanding the system. (3) Thus, the success of student use of MacMan was totally dependent on their interaction with a knowledgeable and skilled instructor.


Our response to this failure of a "naked" simulation (with no actual teaching incorporated in the program) to produce much learning was to produce Heartsim (Rovick and Brenner, 1983). This was a PLATO program made up of two components. The first component was MacMan, the mathematical model. The second component was a "teaching" module.

This didactic component was notable for two things. We realized that we were only really interested in assisting our students to make QUALITATIVE predictions (increase/decrease/no change). As a result we introduced the predictions table as an interface to collect student input, organize the students' thinking about the problem, and to provide the basis for providing feedback about student errors. The didactic component also defined a set of experiments (procedures) for the students to carry out. The protocol for the students was to (1) select a procedure, (2) make predictions, (3) correct logic errors, (4) correct relationship errors, (5) see the results of the procedure plotted and displayed as a table (both carried out by the MacMan component), and (6) compare the students' predictions with the actual results (accomplished by storing the correct predictions - not by looking at the output of MacMan) and provide text feedback to correct the students' lack of information and/or misconceptions.

Heartsim continued to provide the student with an opportunity to design and run their own experiments, although relatively few students were observed to do this.


CIRCSIM is a BASIC program for DOS machines (Rovick and Michael, 1986). Its development was prompted by the fact that few of our medical school colleagues had access to PLATO and it was therefore difficult for us to share our program or get peer review of it.

In translating Heartsim into CIRCSIM we realized that we were making no use of the mathematical model that was such a major part of Heartsim. That is to say, the most effective teaching was being generated from the stored correct predictions for each procedure, not from the quantitative outputs generated by the model.

Thus, in CIRCSIM we store not only the correct predictions but also the limited data needed to display the results of each procedure. There is no mathematical model in CIRCSIM, although all the results displayed by CIRCSIM come from MacMan.

We also made extensive changes in the error checking, including checking for multi-prediction error patterns, and the feedback available for the students was improved. Finally, we provided additional guidance to the students in the form of a "hand-hold" procedure that takes them through the solution to a procedure in a step-wise fashion and then concludes with the development of a general model for a homeostatic, negative feedback system and a kind of algorithm for solving problems about such systems.


Both programs assume that the student have acquired the facts and concepts about the baroreceptor reflex from their reading, attendance at lecture, and participation in the other problem- solving exercises scheduled in the course.

The programs have been used in regularly scheduled computer laboratory sessions in which instructors are present (almost always JM and AR) to interact with the students. We have thus spent over ten year doing one-on-a-few (groups of 2-3) tutoring with students attempting to solve problems about the baroreceptor reflex.

Since such problems also appear on the examinations we write, we have also had considerable experience with one-on-one interaction with students preparing for their exam or wishing to discuss it after the fact.


By the late '80s we realized that while CIRCSIM was an effective learning resource for our students, there were many kinds of errors and misconceptions that it was unable to remedy, in part, because it could not detect their presence.

This conviction arose from the experience of tutoring student using CIRCSIM and observing that even when few if any errors were made, conversation with the students nevertheless revealed significant problems with their understanding of the baroreceptor reflex.

It seemed to us that the solution to our problem would be the availability of a program that could hold a conversation, a dialogue, with student while they were engaged in solving a CIRCSIM problem.

Frank Naeymi-Rad put us in touch with Martha Evens (his former graduate advisor), and the rest is now documented in the theses, presentations and papers listed on this Web site.


Dickinson, C. J., C. H. Goldsmith and D. L. Sackett. 1973.
J. Clin. Computing, vol. 2, pp. 42-50.

Rovick, A. and L. Brenner. 1983.
"HEARTSIM: A Cardiovascular Simulation with Didactic Feedback," The Physiologist, vol. 26 no. 4, pp. 236-239.

Rovick, A. and J. Michael. 1986.
"CIRCSIM: IBM-PC Computer Teaching Exercise on Blood Pressure Regulation," Proceedings XXX Congress of International Union of Physiological Sciences, Vancouver, p. 138.