home articles

Real Time Feedback: Is it real?

Some equipment manufacturers are claiming that their product is the only one on the market having "real time" response. It is implied that digital filters are the only way to achieve fast response times. The following comments are intended to put this situation in perspective.

Effect of Feedback Delay on Training

Feedback delay has a significant effect on EEG training. If the feedback delay is sufficiently short, the user will associate transient emotional and attentional states with the feedback signal. It is not necessary to understand the process at a cognitive level. This type of feedback training is an experience. Learning is fast and permanent. Subjective reports are that fast feedback has a feeling of presence while slow feedback seems more detached.

Defining Feedback Delay

We define feedback delay as the time period between an electrical brain event and the perception of feedback change by the subject. The instrumentation delay in the filter system is only part of the total delay. Other factors such as graphic displays, sound, and effects of artifacts play an important part in the training loop.

Filters

The purpose of filtering is to isolate and make available some subpart of the total or raw EEG signal. For instance, we might use a filter to pass and display only those frequency components that lie between 4-8Hz which we refer to as theta. The process of filtering, however, distorts the signal in ways other than those intended. The signal is delayed in passing through the filter and arrives at later point in time. Effects such as phase shift, bandwidth, and ringing can change the appearance of the waveforms. Often, it is possible to reduce one of these distortions at the expense of increasing another.

A number of filter types and implementations are available to the engineer. While the choice of filter type is a complex issue, the following general information may be of help in understanding instrumentation specifications:

Analog filters: This the term analog is generally used to describe a process where the signal is handled in a continuous stream. The real world is basically analog. Temperature changes take place on a smooth continuum. An analog filter uses electronic hardware circuits to perform the filtering operation. The filter characteristics are designed into the filter by the engineer and are fixed in nature. The digitally tunable analog filter is variation of the fixed analog filter, which allows computer control over the filters characteristics.

Digital filters: Digital systems operate by converting analog information into samples represented by digital numbers. The computer then processes this sequence of numbers according to the instructions of a program (software). A digital filter is simply a program which performs mathematical operations on a stream of digital data. These are the same mathematical equations used to design analog filters and in a sense simply emulate the analog filter in software. Digital filters have the advantage of not requiring hardware which saves on cost. Of course the load is transferred to the computer.

There is no inherent speed advantage between the analog and digital approaches to filtering. Filter delay is more a function of the desired tradeoffs chosen. In general, a narrow bandwidth results in increased time delay. The digital FFT filter breaks the signal into a series of narrow sections each having a typical bandwidth of 1 Hz. A 4-8Hz FFT filter simply adds the 4 through 8 Hz sections to make a wider band. The longer delay time of the filter, however, is determined by the 1 Hz internal bandwidth. The same 4-8 Hz filter implemented as analog or digital emulation will have a much shorter delay. Note, however, that there is no clear difference between the analog and digital approaches. The nature of the EEG signal also puts a limit on filter delay. A 5 Hz theta wave has a time period of 200 milliseconds. Contrary to the claims of some, there is no way to detect a wave in the presence of other waveforms until at least some reasonable part of it has occurred. This limitation cannot be circumvented by increased computer speed and/or sampling rate.

Feedback Displays and Sound

The method used to provide feedback to the subject is a part of any consideration of feedback delay. Until the subject perceives a change (consciously or unconsciously) there has been no feedback. Several things contribute to this effect.

A computer monitor scans new information to the screen every 16 milliseconds. (can be as low as 10mS on some monitors). This puts a lower limit on visual feedback delay.

Visual information must impact the subject’s senses in some way to provide feedback. Scrolling waveform displays provide immediate information only at the point where the graph is being updated. This takes the form of one moving dot (pixel) on the screen. The remainder of the display shows the last 2-8 seconds of data which cannot be modified by the subject. Even if this information shows trends, it is not part of "fast feedback". Bar graphs, box graphs, intensity displays, etc. are much more effective in providing a focal point for the subject’s attention.

Sound feedback must also provide perceptible indications of change. A sound which beeps at 0.5 second intervals to indicate criteria extends feedback delay to 1/2 second. Sound must also change sufficiently to be noticed. Sound is not limited by the screen update time.

Conclusion

Feedback instrumentation is governed (and limited) by the laws of physics and electronics. At Focused Technology, we make no claims to proprietary or patented technology. Our equipment is based upon solid engineering principles available to anyone else in the industry. We simply make compromises based upon our knowledge of the needs of EEG feedback and the nature of the signals involved. We make our specifications available and describe the methods used to make the measurements.

Our concern is that we are increasingly encountering clinicians who have paid excessive prices for systems with no clear advantage over the F1000. The technical nature of the issues involved leave the clinician vulnerable to marketing hype. A writer in a professional electronic engineering publication recently wrote "..marketing has never been hampered by reality".

Feel free to contact Frank Deits if you have questions or comments.

home articles top