General Method

A.1.1.1 Experimental Apparatus and Stimuli

The experimental apparatus consisted of a standard 666 MHz Pentium III computer with 256 Meg of RAM, running the MS Windows 2000 operating system, using standard input devices (serial mouse and ps2 keyboard), and a 17” Hyundai cathode ray tube monitor with a refreshment rate of 80 Hz. The contact of the right mouse button was fitted with two additional wires so that a mouse press could be simulated by the auditing computer.

A custom program was scripted in Inquisit software (Draine, 2003). This software package was selected because it has been designed to have high timing accuracy. It accesses the peripherals using Microsoft directX drivers. This means that, at critical times, when stimuli are being displayed and response times are being measured, the program takes full control of the input and output devices (mouse and monitor) which are then no longer accessible by any other program that may be running in the background. The program also increases the number of times the mouse is polled for responses to the maximum frequency, can trigger external devices (i.e. the computer auditing the timing accuracy) via the parallel port, and only allows activity of processes necessary to the stability of the operating system.

Stimuli consisted of a white square with sides of 2 cm presented in the top left corner of the screen on a black background. Since the stimuli were used to trigger the auditing computer, and since the screen refresh starts in the top left corner and finishes in the bottom right corner of a cathode ray screen, it was important that the stimuli be presented in the top left corner of the screen so as to indicate the beginning of the screen refresh.

During each trial, a stimulus was presented for 50 ms, and a mouse response was expected from the auditing computer to indicate detection of the stimulus. The trial terminated either when the mouse response was received or after a time-out period of 400ms. Each trial was followed by an intertrial interval of 500 ms.

A.1.1.2 Auditing Apparatus

The auditing apparatus consisted of a 1.6 Mhz Pentium IV computer with 512 Meg of RAM, and fitted with a National Instrument analog/digital data acquisition card and connection box. Two inputs were connected to this system. One came from a solenoid probe sensitive to light signals, which was attached to the screen of the experimental computer and served to detect the presentation of stimuli. The other came from the parallel port of the experimental computer and was used to read signals indicating significant events (stimulus presentation and end of trial) of this system. The connection box also provided one output used to simulate a mouse response and trigger the experimental computer. A customised program was developed to control this hardware using National Instruments’ LabView software. This program was designed to detect the stimulus onset and presentation time, trigger a mouse response after a fixed period of time, detect the data sent from the mouse to the computer, and measure the total trial length by detecting parallel port signals from the experimental computer sent at the beginning and end of each trial.

A.1.1.3 Design and Procedure

In all experiments, a stimulus was generated by the experimental computer while at the same time, a parallel port signal was send to the auditing computer. This was done to measure the length of the delay between stimulus initiation and its appearance on screen. Unlike the parallel port signal, which is transmitted almost instantaneously, the screen stimulus needs to be buffered and synchronised with the screen refresh. If one of these events was not adequately controlled by the software, the difference between parallel port signal and appearance of the stimulus on the screen would be substantial. If this were the case, this would translate in a significantly shortened stimulus presentation time.

Each stimulus was presented for 15 refresh frames lasting 11.76 ms each for a total duration of 176.47 ms. The appearance of the stimulus on screen was detected by a solenoid detector attached to the screen and connected to a signal processing box, itself connected to the data acquisition box of the auditing computer. This signal started two timers in the customised software of the auditing computer. One timer measured the length of different events produced by the experimental computer (e.g. total trial length, stimulus-to-mouse response length, stimulus presentation length). The other timer was used as a response delay, and simulated a mouse response by triggering the right mouse- button switch of the experimental computer after 300 ms.

The experimental computer’s mouse was also wired to provide information about the data it sent to the computer. These wires could be connected either to the data acquisition box to determine the internal delay of the mouse, or to an oscilloscope to measure the shape of the data signal.

Once the mouse was triggered, the signal received by the experimental computer terminated the current trial, which in turn triggered a second parallel port signal to indicate to the auditing computer that the trial was terminating. This signal stopped all the active counters in the auditing computer.

A new trial was then started after a 500 ms inter-stimulus interval. The Inquisit software recorded stimulus lengths and trial lengths to file.

The auditing computer was fitted with three counters and therefore, had the capacity to record a maximum of three durations. As more than three measures needed

to be recorded to audit the experimental apparatus, a number of experiments were run to measure different variables of interest. All these experiments shared the procedure described above. Additional manipulations, specific to each experiment, will be described in the next sections.

A.1.2 Experiment 1

This experiment was designed to assess the delay between the start of a trial and the appearance of the stimulus on-screen, and to measure the delay between mouse button trigger and the mouse digital data transfer to the computer (mouse buffer).

This second variable is particularly important because input devices, such as computer mice, do not forward a response directly to the computer. Instead the button press is stored in a memory buffer in the mouse and transferred only when the computer polls the mouse for responses. The computer polling frequency is controlled by the operating system under normal circumstances and can be adjusted. However, the software, when MS directX is used, can adjust this parameter to allow for the fastest possible scanning of the mouse buffer, thus improving the timing accuracy of the mouse response.

A.1.1.4 Procedure

One timer measured the duration between detection of the first parallel signal (start of trial) and detection of the visual stimulus on-screen. The second timer measured the duration between visual stimulus detection and the mouse button trigger. This was done to ensure that the length of the delay of the auditing computer before initiating a mouse response was accurate. The third timer measured the duration between closure of the mouse switch (mouse press simulation) and detection of the data transfer from the mouse to the computer. Simultaneously, an oscilloscope also connected to the mouse data cable recorded the length and the shape of the data sent to the computer and provided a graphical snapshot that could be viewed on a computer. This was done because the experimental computer would not be able to pass the mouse response to the software before the complete data packet was read. Therefore, the length of the data packet would add a fixed delay to the response time recorded by Inquisit which would need to be included in the timing audit. One thousand trials were recorded in one testing session.