Limitations of high-level microcomputer languages in software designed for psychological experimentation

Abstract

The execution times of microcomputer high-level-language commands can be long enough to be of concern in experiments in which precise timing is a consideration. The problems in developing standard BASIC timing routines are addressed. A technique for using the Model III TRS-80-real-time clock to calibrate BASIC timinig routines is described, and representative execution times of selected commands are reported. It is concluded that high-level languages are too slow and that execution times are too variable for critical timing in experiments. On the other hand, machine language programs can provide the needed precision and control.Timing with millisecond precision is a major requirement for microcomputers used in research. The timing of stimulus exposures, interstimulus intervals, and response latencies can be accomplished by a software timer or by having the computer input signals from an external clock. Each technique has advantages and disadvantages. An external clock is reliable, but it is also an additional expense and must be interfaced with the computer. A software timer requires no additional equipment, but it can be sensitive to tampering. Moreover, timers written in high-level languages (e.g., BASIC or PASCAL) are too slow for millisecond precision. This has led some researchers to rely on external timers for critical timing. Regardless of the timer used, however, reliable event timing also depends on the execution time of program commands, whether they are written in a high-level language or in machine language.For purposes of this discussion, assume that an experiment consists of a series of discrete trials on which a stimulus is displayed, timed, and then masked. Also, a latency timer starts at stimulus onset and stops when the subject responds. Millisecond timing is required for stimulus exposure, interstimulus interval, and response latency: Either an external timer or a machine language software timer may be used for this.A detailed analysis of trial events shows that a reliable timer is not all that is needed to assure that program timing fits the intended specifications. With a software timer, the following operations must be performed: (1) display the stimulus, (2) start timing response latency, (3) time stimulus exposure, (4) display the mask, (5) monitor the subject's response, and (6) stop and read the latency timer. With an external timer, the operations are these: (1) read the clock for the start

Software timing of events in cognitive psychology experiments

“…A software clock can be created by executing a delay loop in a higher level language such as BASIC (Rosner, 1978). Although this method is convenient, it suffers problems in coarseness of resolution and timing variability and thus is oflimited usefulness (Dlhopolsky, 1983;Price, 1979). In general, it would appear to be difficult to write a precise software clock in a higher level language; Methods 2,3, and 4 should be written in assembly language.…”

TREC: A tool kit for programming cognitive experiments in Applesoft BASIC

“…First, it is usually necessary to implement a timer for measurement of response latencies and stimulus exposure duration (Dlhopolsky, 1983b). Second, programming in BASIC is usually inadequate, because it is difficult to synchronize the program with the videoscan of the CRT and to measure the time (Dlhopolsky, 1983a).…”

Visually presented verbal stimuli by assembly language on the Apple II computer

“…The high-level program is run under interpretation or compiled into machine code which is executed directly. Under interpretation, the execution time for a given program line can be quite long and variable (Dlhopolsky, 1983). When a program is compiled, This work was supported in part by BRSG Grant RR-07062 awarded by the Biomedical Research Support Grant Program, Division of Research Resources, National Institute of Health.…”