Frequency-Following Response (FFR)

Written by Travis M. Moore
Last edited 22-Jul-2020

A Cochlear Microphonic in the Brainstem?

As discussed in the lesson on electrocochleography (ECochG), the cochlear microphonic (CM) is a cochlear potential (mostly from the outer hair cells) that "follows" the "shape" of a low-frequency stimulus. In other words, if we present a 500-Hz pure tone, the waveform of the recorded CM should be a roughly periodic wave at 500 Hz. The CM lets us see the cochlea responding to sound. Similarly, the frequency-following response (FFR) also mimics the "shape" of the stimulus, but arises from neurons. The typical latency of the FFR is around 6 - 8 ms; so, where would you guess the neural generators are? Think about the latencies and proposed sites of origin of the waves of the ABR. By the time waves show up around 6 - 8 ms we're likely picking up activity from around the inferior colliculus, which is in the midbrain at the top of the brainstem. So the FFR is a response that follows the stimulus like the CM (in fact, the CM contributes to the FFR response), but arises from rostral (upper) brainstem neurons.

Potentials of the FFR

Two main potentials make up the FFR response. One has a latency of less than 1 ms, and most likely comes from the cochlea (this is likely the CM). The other potential has a latency of around 6 - 8 ms, and likely arises from the midbrain (upper brainstem). So the earlier source is cochlear in origin, and the later source is neural. Even though both potentials follow the waveform or envelope of the stimulus, the FFR reveals information about how the brainstem processes sound that the CM cannot give us. The fact the FFR exists at all tells us neurons in the midbrain are able to phase lock (fire one every cycle). So if the interesting thing about the FFR is the neural component, yet again, we oftentimes want to remove the CM. The poor CM can't seem to catch a break...

Out of work CM
FIG. Background photo modified by Travis M. Moore (2020).

Studies examining single neurons in the brainstems of animals has revealed that the neurons of the auditory nerve (AN) phase lock the best, and neurons higher up in the brainstem phase lock comparatively worse. (See the note below for a brief discussion on this principle.) That means the cochlea/AN can lock on to higher frequencies. Remember higher frequencies have smaller wavelengths and can fit in more cycles per second than low-frequency, longer-wavelength frequencies. The phase-locking cutoff in several single neuron studies was around 1000 Hz, and accordingly, the neural part of the FFR becomes overshadowed by the CM at frequencies around 1000 Hz.1 More evidence the neural FFR comes from the rostral brainstem.

Ok, so what does the neural FFR look like? If the stimulus is less than around 1000 Hz, the short answer is: what does the stimulus look like?

Normal auditory brainstem response waveform morphology
FIG. The FFR response to several frequencies.

Interpreting the FFR

Because the FFR closely matches the spectrum (frequencies) of a stimulus, it is often most useful to consider the actual spectrum of the FFR; that means looking at the response in the frequency domain.

Recording the FFR

Because the stimuli for the FFR can vary wildly, it is not prudent to expect a single set of parameters to work in every situation. For example, you will have to consider the frequencies of interest before setting your filters, and use a time window that is appropriate for the length of the stimulus, PLUS some "padding" (around 10 - 20 ms) to make sure you capture everything. Below are some example parameters to get you started using single tones.

TABLE 3. Vertical electrode montage for the neural FFR. This montage is appropriate for tones below 1000 Hz.
Electrode Site
Non-inverting (active) Cz (vertex)
Inverting (reference) Ipsilateral earlobe (A) or mastoid (M)
Contralateral earlobe*
A cervical site*
Ground (common) Fpz (low forehead - center)
TABLE 3. Horizontal electrode montage for the neural FFR. This montage is appropriate for tones between 1000 Hz and 2500 Hz.
Electrode Site
Non-inverting (active)** Ipsilateral earlobe (A) or mastoid (M)
Inverting (reference)** Contralateral earlobe
Ground (common) Fpz (low forehead - center)
TABLE 5. Example tone-evoked neural FFR amplifier parameters.
Parameter Value
High-pass filter 30 Hz
Low-pass filter 3000 Hz
Gain 100,000 - 200,000
TABLE 6. Example tone-evoked neural FFR stimulus parameters.
Parameter Value
Stimulus Tone or tone burst
Duration ~10 cycles for continuous tones
~40 ms for tone bursts
Intensity Between 40 and 70 dB SL
Disappears around 30 dB nHL
Polarity Rarefaction
Rate Varies based on stimuli.
~12 /s for tone bursts
Sweeps 2000

Clinical Applications of the FFR

Neuronal Indicator of Brainstem Speech Processing

Content not yet available.

Neuronal Indicator of Cochlear Distortion Products

While discussing ECochG we learned that the CM originates mainly from the outer hair cells (OHCs), which means the CM is sort of an electrophysiological correlate of distortion product otoacoustic emissions (DPOAEs). Well, it turns out the neural FFR is yet another correlate to DPOAEs.

You'll recall that DPOAEs are a sign of cochlear nonlinearity. If you're unsure of what cochlear nonlinearity is, you might want to review your psychoacoustics notes. Very briefly, in a linear system we get out what we put in - and nothing more. Sure the output might be consistently larger or smaller, but there really won't be any surprises in linear systems. An example would be a really well-built cell phone. You speak into it (the input) and the signal is transmitted to the person on the other end without any additions. The result? Your voice sounds nice and clear.

A nonlinear system does not play by these simple rules. It's possible to see things in the output that we never put in the nonlinear system in the first place. A great example of this is an actual cell phone. When you speak into the phone your voice sounds clear, but the output - that's a different story. Because of the cheap components in most cell phones at the time of this writing, your voice comes out a little distorted or static-y on the other end. That extra "noise" wasn't put into the nonlinear system, but it was added along the way.

So, what does this have to do with the inner ear? Well, the cochlea is a nonlinear system. So if we put two tones into the ear simultaneously, we don't just get a nice, simple interaction. Instead, the cochlear response includes its own distortion. This is the essence of what DPOAEs measure. The distortion product of the two tones going through the nonlinear cochlea. More specifically, we put two tones in, and we're listening for a third (i.e., the cubic distortion product).

DPOAEs measure OHC function, so they are a cochlear/peripheral test. The neural FFR, however, also picks up a distortion product that has made it all the way to the upper brainstem (i.e., the quadratic distortion product). (By the way, cochlear nonlinearity is not bad, which is a. why it exists, and b. why it's preserved in the brainstem and cortex.)

So how do we use this in the clinic? Good question. There hasn't been a lot of work done in this area at the time of this writing. One potential use is that the FFR is sensitive to low-frequency distortion products, while DPOAEs typically start to sink into the noise floor at frequencies below around 1000 Hz. This is a wide-open area for anyone with a budding interest in research!

Test Your Understanding

The rostral (upper) brainstem, maybe around the inferior colliculus in the midbrain.
Answer Three


Batra, R., Kuwada, S., & Maher, V. L. (1986). The frequency-following response to continuous tones in humans. Hearing Research, 21(2), 167-177.
Picton, T. W. (2010). Human auditory evoked potentials: Plural Publishing.
Atcherson, S. R., & Stoody, T. M. (2012). Auditory electrophysiology: a clinical guide: Thieme.
Leave a Comment: