An integrative approach to uncovering the neural basis of pitch perception.

After a sound wave enters your ear, the activity of hundreds of thousands of neurons throughout your brain transforms this signal into a perceptual experience. The tonal quality that we ascribe to sounds, known as their "pitch", is one of the most important features of this experience. Pitch perception allows us to recognize a familiar musical melody, the low growl of a lion, or the inquisitive tone of someone's voice. Many animal species also use pitch to communicate and interpret their world. For instance, mother monkeys have been shown to raise the pitch of their voice when calling their young, just as human mothers do when speaking to their children. As we age, our ability use some types of pitch cues declines, even in individuals with normal hearing thresholds. Therefore, an understanding of how brain cells compute pitch, and how age and experience can change these processes, may help to improve hearing in elderly people.

While our ability to distinguish a low note from a high note comes effortlessly to us as listeners, it has proved a great challenge for neuroscience to fully understand how our brain accomplishes this feat. Functional imaging studies in humans suggest that certain parts of the brain may be specialized for pitch perception, but there is still much debate over where exactly the brain's "pitch center" is located. Moreover, these brain imaging techniques do not have sufficient resolution to tell us how nerve cells in the brain represent the pitch of a sound. Experiments that measure the activity of individual brain cells in animals are therefore essential to answering to this open research question. Our research at the University of Oxford will use an innovative combination of modern neuroscience methods to uncover the neural basis of pitch perception.

It is unclear how pitch perception in many animals relates to that that of humans. We will therefore begin our studies by comparing the physical properties of sound that humans and our animal model, the ferret, use to determine its pitch. We will train the animals to distinguish low- from high-pitched sounds on a behavioural task, and human volunteers will be trained on a similar task. Because the nerve cells in the ear are better understood and simpler than brain cells, we can build computer models that predict how nerve cells in the ear will respond to different sounds. By comparing these model predictions to real behavioural performance, we will determine how animals and humans use sound features to make pitch judgments. To discover how neurons in different parts of the brain represent sound features that are relevant to pitch perception, we use microelectrodes to measure the activity of nerve cells in the animals while they perform the pitch judgment task. The microelectrodes will be implanted under surgical anaesthesia, so that the animals do not feel pain. By recording brain activity in the animals as they are trained to derive pitch from different sound properties, we will reveal how the brain's representation of pitch can adapt through experience. Finally, we will use new, laser-based microscope technology to make videos of the activity of large numbers of brain cells while animals are listening to sounds. These experiments will allow us to map out how pitch-selective nerve cells are distributed across the surface of the brain, with a density and resolution that have never before been possible. The results will indicate whether a pitch center exists in the brain, and, if so, where it is located. Together, these experiments will significantly advance our knowledge of the brain processes that allow us to follow the melody of our favourite tune.

Pitch is one of the most salient and behaviourally relevant perceptual features of sound for humans and animals. Poor encoding of temporal pitch cues is thought to underlie the communication difficulties experienced by elderly individuals, but little is known about how animal models make use of temporal pitch cues or how these cues are encoded by the auditory cortex. The studies proposed here will aim to answer these fundamental questions by combining the strengths of behavioural, computational, single-cell electrophysiological and single-cell imaging techniques. Firstly, we will compare how ferrets and human listeners use temporal and resolved harmonic cues on a pitch judgment task, and this performance will be compared to the predictions of auditory nerve models. Next, we will measure how neural spike trains represent pitch cues throughout the ascending auditory system (including the inferior colliculus, primary and secondary auditory cortical fields). By measuring neural responses in animals as they perform the pitch judgment task, we will investigate how training on specific pitch cues may change this neural code. Trends in our previous datasets suggest that a "pitch center" may exist on the low-frequency border of primary and secondary tonotopic auditory cortex, but this possibility requires direct experimental testing. We will use a combination of microelectrode recordings and 2-photon imaging of calcium dynamics to determine if a pitch-selective region exists in ferret auditory cortex. The temporal precision of microelectrode recordings will be used to identify the spiking codes for pitch in this region, while the superior spatial resolution of 2-photon imaging will allow us to densely sample the tuning properties of neighboring neurons in this region. The 2-photon imaging experiments are ideally suited to clarify whether the pitch of complex sounds is mapped across the auditory cortical surface.

Our research investigates the sound properties that animals and humans use to determine the "pitch" (i.e. the tonal quality) of a sound, and determines how activity in brain cells carry out this process. The interdisciplinary nature of this research means that it will impact upon a wide range of academic fields, which are fully described in Academic Beneficiaries. Significantly advancing just one of these areas would provide a huge step forward for neuroscience.

Our integration of computational modeling with behavioural and physiological studies aims to promote future links between mathematical biology and neuroscience. We will also develop methods to efficiently measure the activity of very large numbers of brain cells simultaneously, through multi-channel electrophysiology and 2-photon imaging. These methods allow neuroscientists to answer questions about brain function more quickly and economically. We know of only 2 other labs worldwide that carry out in vivo 2-photon imaging in ferrets, so by developing this technique we will enhance the UK's research portfolio. Together, these modeling and high-yield physiological studies answer society's call for replacement, reduction and refinement of the use of animals in scientific research.

The public has a natural interest in how animals experience the world. Our research offers insights into how our own perception of the musical quality of sounds relates to that of other animal species. This knowledge can help pet owners and farmers communicate with animals, and may guide government regulations that protect animal's natural environments (e.g. does traffic noise affect animals' ability to communicate?). We will develop our computational model of animal hearing into one in which anyone can listen to the sounds in their current environment through "animal ears". We will make this program available for people to use on our website, and if it is popular we will also develop it as an "app" for mobile phones and ipads.

Ageing has a detrimental effect on people's ability to process temporal pitch cues, and this in turn limits their ability to communicate in moderately noisy environments, such as restaurants. One study suggests that this temporal processing impairment may begin as early as middle age, so it is likely to affect the day-to-day hearing and social health of a large portion of society. Our research will provide new insights into the relation between temporal processing and pitch perception. We will elucidate the neural codes for temporal pitch cues, which are likely to be the same neural mechanisms that are disrupted in aging hearing. Our research will help clinical audiologists understand the impact of age-related hearing impairments on pitch perception.

The commercial development of training-based treatments for temporal processing impairments in children have had successfully improved hearing and language outcomes for some people. The hearing of elderly individuals may similarly be improved through training regimes, but this possibility remains to be explored. Our experiments will help to push this area forward, by testing how training in adult ferrets can improve pitch discrimination thresholds and how it may alter the neural code for pitch.

As predominantly temporal pitch processors, ferret experiments could be a useful model of pitch perception in cochlear implant (CI) users. Due to the technical limitations, CIs provide highly degraded spectral resolution to the auditory nerve, so CI users are more reliant on temporal cues than healthy listeners. Unfortunately, the temporal signals currently provided by cochlear implants are also very crude, and so pitch perception in these listeners is severely impaired to the point where they cannot hear the melody of most music. Therefore, understanding how the brain makes adaptive use of temporal and spectral pitch cues will help companies to design better cochlear implants and hearing aids.