The Department of Psychology uses a system called BEST in order to record and process EEG data acquired from test persons. To these persons pictures and sounds are presented to find out which cerebral regions are stimulated and in which way human senses can be deceived. E.g., the person sees a rapid sequence of pictures which suggests a tropical climate and one frame of this sequence presents somebody drinking a refreshing beverage. The aim of this scenario is to find out whether humans subsconciously are capable of processing these (hidden) one frame pictures.
The existing system only provides single tunes and colored words shown in text mode as stimuli.
Therefore a system (STIMULAT) is to be developed with the following main characteristics:
Because the BEST- system has been designed as a stand alone solution a means of communication with another PC ha s to be developed. BEST uses a special interface consisting of 2*16 parallel digital lines which can be used either as input or output. In order to communicate we created an interface card consisting of a serial interface and a programmable IC that maps signals coming from BEST to standard V.24 signals and vice versa and generates ASCII characters (needed to transfer the filename that contains the test information). The communication between BEST and STIMULAT via the interface is unidirectional; apart from sending the filename BEST used two other lines for indicating a program-reset and switching to the next picture/sound.
Each of the events noted above triggers an interrupt. In case of a switching event the interrupt routine waits for the next vertical retrace signal of the CRT controller to ensure fully shown pictures. This is crucial, because only partially shown pictures (one frame) would clearly result in a different stimulus and therefore ruin the test.
To guarantee a fast loading routine the pictures used as source (Windows BMP pictures with 800x600 pixels and a colour depth of 24 bit) are transferred to an image that is colour-reduced to 16 bit and its picture information is already mapped for the specific video card and video mode. This is done prior to the actual stimulation, so that the calculated video image only needs to be copied to video memory, which guarantees fast loading. It is now possible to load up to five different picture samples per second if needed. We used the Matrox Millenium video card with 4 MB of memory, which allowed us to store up to four pictures a time.