Poppy Crum, PhD, might be hard to categorize, but her interests are not as disparate as they might seem. She began playing violin when she was three years old. Her fascination with sound and the way it is perceived — in particular a phenomenon called absolute pitch (or perfect pitch), wherein the names of musical notes are associated with hearing changes in frequencies — helped lead her to the study of neuroscience.
Today, Dr. Crum conducts auditory research at Johns Hopkins. She studies the neural circuits involved in hearing in a complex acoustic environment.
“When we hear, the problem presented to the auditory system is quite complex,” she says. “For example, in a noisy environment, sound waves coming from many sources – a dog barking, multiple conversations, a car driving past -- combine to create a complex waveform that reaches your ear. Somewhere in the auditory pathway, the system has to figure out which frequencies belong with which sources, where they are, and how loud they are. The system also has to know which acoustic information and visual information belong to the same object, and in many instances fill in information that isn’t even there to enable successful navigation through the environment.”
Her work focuses on understanding how the auditory cortex is involved in solving this problem.
In rehearsal for the Baltimore Baroque Orchestra. Dr. Crum studied violin at the University of Iowa and the conservatory of music at McGill University.
Originally from Iowa City, Dr. Crum received a degree in violin performance studying at the University of Iowa and the McGill conservatory of music in Montreal. (Listen to Dr. Crum playing part of the 3rd movement of Shostakovitch's 7th String Quartet.) Her interest in hearing and audio grew while having extra jobs during music school working as an audio engineer in various venues, as an announcer for the local NPR radio station, soldering circuit boards for one of the original designers of the synthesizer, and consulting in the design of (and performing in!) musical toys aimed at stimulating cognitive development. Her interest in science grew while working in various laboratories at McGill.
She received a master's degree in psychology with a focus on auditory perception and cognitive psychology. Continuing her research on the auditory system and neuroscience, she completed her PhD at the University of California Berkeley. A post-doctoral fellowship in biomedical engineering -- studying the cortical neurophysiology of the auditory system -- brought her to Johns Hopkins in 2005 where she has remained ever since and is now a member of the research faculty.
She is also on the faculty at Johns Hopkins' Peabody Conservatory, where she lectures on psychoacoustics, and where she performs as a musician. Dr. Crum is often invited to speak and share her knowledge of hearing with the music and audio industries.
The Lost Coast trail in California near the Oregon border goes for about 26 miles. One afternoon, Dr. Crum ran the entire length. She has won many trail runs including the Mt. Diablo Double race in Calif., 50 km long with 9,000 feet of vertical gain.
Q&A with Dr. Crum
Hi everyone, I look forward to your questions. I've put links to a few well known auditory illusions. Take a listen if you get a chance and we can talk more about theories of why and how they happen. . . .
Finally!!! Something I actually understand!!!! (Or so I think...) Welcome to Cogito, Dr. Crum! What do you think of Oliver Sack's work, especially Musicophilia?
Glad you liked the illusions! Have to admit I haven’t read Musicophilia, but am familiar with many of the syndromes Sacks discusses. He is a master of introducing and describing the most indescribable human neurological conditions. The Man Who Mistook His Wife for a Hat was supplementary reading for one of my first neuroscience courses, and it is hard not to find conditions of visual agnosia or prosopagnosia (the inability to recognize particular objects or faces) fascinating.
It was also surprising to find out how many people experience some form of many of these conditions that seem so bizarre and far out. When I was at Berkeley colleagues began soliciting for prosopagnosic subjects online (this was a new thing to do – believe it or not!) and soon had many with varying degrees of the condition. Far more than they had ever planned. . . . Similarly, I believe amusia is much more prevalent than most realize.
Sacks has provided an important service that crosses over the science border – he puts a name and gives a voice to the perceptual experiences that individuals (and/or those who know them) experiencing a particular condition might not be able to describe without identifying through Sack’s text. To be fair, many of the syndromes he describes have been known for many years in the science community (-- known, but FAR from understood), however, he brings them to a different level of understanding that can have a great benefit. It’s also important to point out that the conditions he tends to describe are often the most quirky and perplexingly interesting.
As a musician, your research makes a lot of sense to me. Can you apply your findings about tuning to vocal music as well? It seems like some people are just born with good internal pitch. Can you train the ear of a singer to have perfect pitch?
Nice question. First, it is a good idea to differentiate what we mean by “absolute” pitch (which is the same as “perfect pitch, but a more accurate description) and “relative” pitch. Having a good internal representation of pitch can easily come from a relative representation. The distinction made with absolute pitch is that frequencies are mapped to particular note names, and that the experience of hearing and knowing these names when listening to sound is experienced much like seeing color when you look outside.
Dr. Crum climbing at Phantom Spires in Lake Tahoe, one of the beautiful outdoor settings where she has enjoyed rock climbing.
Absolute pitch comes in many varieties. For instance, it is not uncommon for a musician who doesn’t have absolute pitch to any encountered pitched or semi-pitched sound (e.g. the hum of a refrigerator) to develop absolute pitch specifically for the instrument they play and others in a similar class. Studies have shown people can be trained to have extremely good pitch memory. In one instance listeners trained repeatedly in listening with tuning forks learned to associate particular note names with each fork’s frequency and were able to make accurate identifications on a later date. This was interesting – it showed learning of associations that were not made under listening conditions of relative pitch and previously not believed to be easily learned. However, in that instance the sample set was quite small, and memory strong.
However relevant such learned pitch association may be, it can be assumed that absolute pitch is a much different process where any encountered stimulus - regardless of familiarity - is processed with a strong named association. The evidence for having this type of pitch processing points towards an existing genetic predisposition and exposure in formative years to either music training and/or tonal languages. Nonetheless, as a vocalist you also develop a very strong proprioceptive memory for your production of individual pitches that may enable both production and experience of pitch that is in many ways similar to absolute pitch. What have you noticed?
I was wondering two things about the auditory illusions. First, how does the second illusion work? When I listened to the video you linked to, I only heard "ga." Then, I listened to a different version where the person was mouthing more slowly, and only heard "ba." Any ideas on why that might be? (Other than my ears and eyes just not liking to work together.)
Also, I thought of a real-life example of the McGurk effect. If you listen to a whispered conversation while looking at the speakers, it's easy to understand. But let's say you don't look at them: then it's really hard to hear! The funny thing is, though, people always seem more inclined to think you're eavesdropping if you look at them. Is it possible that we subconsciously know that our brains work this way, and interpret people looking at us accordingly?
Interesting thoughts! The McGurk Effect is a great example of how we perceive a non-veridical world (stimulus to perception is not 1:1). In the video, the speaker is moving his mouth as if to say “ga” while audio has been tracked to the video of him saying “ba.” Many people perceive “da” - an illusory combination of the two. In this case the information coming from the visual and the auditory components don't align with our experience of the world (e.g. we have seen and heard many people saying "ga" and "ba" leading to probabilistic expectations of these visual and auditory pairings). Here, someone is saying what looks like "ga" but sounds like "ba" and our brain tells us it can't be -- the resolution is that it must be "da." The visual speech component of "da" is much more similar to "ga” than “ba,” and the auditory component fits with small variance.
If you experience the illusion, it is difficult to override; however, not everyone does. For example, presumably a non-native English speaker who has acquired a very different set of probabilistic occurrences regarding their native language, and more importantly lacks those that a native English speaker would possess, would not experience this phenomenon as strongly with this particular stimulus set. In summary – we resolve incongruous information as best we can to fit with the most probable outcome - illusions like this are the result of this happening and getting it wrong.
I suspect in your experiences of the McGurk effect two things happened. First, the perception of “ba” in the slo-mo condition makes sense. Two features binding the auditory/visual information in the normal example are location and timing. Timing is a powerful cue. Our system likes to assume things that happen together belong together. Once you decouple this feature there is every reason for your system to interpret the information as coming from two sources. Once this happens you should experience the actual audio of “ba” as you did.
What is the third illusion supposed to sound like? The web site seemed to indicate that most people hear a continuous increase in frequency, but I didn't hear that at all. I don't know if this is a problem with my headphones, my ears, or my understanding of what was supposed to happen.
In this instance, I suspect it does have to do with your headphones. Try listening over your computer speaker. With headphones, you are taking away the integrated location cue. While obviously your computer speaker is in a different location than the visual – ventriloquism is powerful enough to override that separation. Some headphones may produce a different result simply by subtle design differences causing shifts in the location of the stereo image. Let me know what happens!
What did you hear?
I heard a tone increasing in frequency, then another tone overlapping it and increasing frequency even more. However, I could very clearly hear the second tone coming in over the first. I'm not sure if that was the illusion, or if the illusion didn't work.
Thanks for sharing your perception. For all I will describe what the expected experience would be, but other perceptions are entirely possible - there is never right or wrong - just interesting :)
Listening to the continuity illusion, you hear four brief demonstration. In each, there is a rising tone with a brief gap in the middle. What differs in the four demonstrations is what fills that gap.
- In the first, a noise burst is introduced during the gap. However, the listener hears the tone as continuous, even though the tone is replaced by noise. This is the illusion of continuity.
- The second and fourth examples are identical. In these, no noise is added during the gap, and there is a perceived gap between the two sweeps.
- In the third example, a noise burst is again added; but in this case, the spectral energy in the frequency range that would join the two sweeps has been removed – there is a spectral “notch” in the noise. The perception in this case is of two shorter sweeps and should be similar to the conditions two and four.
The “notched” noise example illustrates the importance of having energy in the frequency range of the missing (gap) information in order for the auditory system to fill in the gap, so to speak. In order to do this, the auditory system has to be stimulated as if the sweep were continuous. What is interesting about the continuity illusion is that we are able to use the energy from noise to experience perception during the gap in the swept tone - as we do in real life, where we often encounter complex stimuli that are rapidly changing in spectral content and must be resolved by the brain.
Does this seem consistent with your experience, or did you hear all four as discontinuous? I am interested if you also heard the first as discontinuous (this is the only condition that should have been perceptually continuous). Perception is at the end very unique, and these are the times when it can be the most interesting :).
Speaking of illusions, what do you make of the recent Nature article that reports that humans hear with their skin (Nature 462|26 November 2009:502-504)? Moreover, when deaf people read sign language, auditory cortex shows signs of activity. And when blind people "read" Braille with their finger tips, visual cortex becomes active. How does polysensory integration occur? Do these phenomena suggest that the principle of "localization of function" must give way to a more distributed organization of function?
What a nice and interesting question. I have seen the paper you mentioned, and while the idea that we integrate tactile and auditory information in forming our perceptual experience of sound is certainly interesting, the paper itself left me a little flat. Scientifically it didn’t convince me, what did you think?
But the general question regarding polysensory integration is both fascinating and critical to navigation and survival in a complex environment. For example, in conditions of auditory/visual integration for an object, such as a crying baby or a tiger rustling leaves as it emerges from the bushes, the perceived location of the baby and the tiger based on the available localization information from each independent modality (audition/vision) can be quite different. In general we have much better accuracy determining where an object is through vision than audition (as long as it is in our field of view). There is considerably more variance in our perceived location of an auditory source in the environment.
To be efficient, somewhere and somehow we need to combine this conflicting information and perceive a single location of the object to best not be gobbled by the tiger. The result is typically that one modality has to win out and dominate our perceptual experience. In this example, most often it is the visual modality -- one might say we live in a constant state of ventriloquism!
An interesting question in these cases of sensory integration is whether one modality is innately dominant or, instead, whether the system is more fluid and simply weighting the more reliable information in a way that is malleable. Many recent perceptual studies have elegantly manipulated the inherent variance of the different modalities and found that sensory integration is nicely modeled using Bayesian algorithms which would suggest the latter.
So, back to your question, things clearly get pretty interesting when we start thinking about how sensory integration will be represented neurally and more so what happens when one sense is truly dominant, as in situations where someone is blind or deaf. The studies you mentioned show the plasticity of the brain in full swing with the dominant modality essentially assimilating the functionality of neurons in cortical areas of the unused modality. For example, in auditory cortex, it is quite common to find, nestled among cells preferring sound, cells that are optimally driven either solely by visual stimuli or preferring some simultaneous combination of both sound and light. This all happening despite clear, well-established, regional modality organization across the cortex indicating that localization of function and a distributed organization are not mutually exclusive neural representations.
In support of this, cortical anatomical tracing has indicated a much more diffuse spread of connectivity from one modality to another, but the role of these connections in the auditory cortex are currently not at all well understood; nonetheless, they are a likely candidate for being involved in the type of neural plasticity that would enable neurons speaking the language of one modality to switch to another.
Making things even more complicated, or interesting and exciting as we all like, a recent behavioral study I found fun and interesting showed that the perception of motion in one modality dynamically impacted the perception of motion in another (Current Biology, Volume 19, Issue 9, 745-750, 09 April 2009). In the study the experimenters induced a well known visual motion after-effect (when a viewer observes a visual grating moving in a single direction for a period of time they experience motion in the opposite direction upon cessation) and measured tactile sensation. The well-known visual after-effect transferred to the tactile modality and vice versa. Pretty cool. In any event, future neuroscience has a lot of discovery ahead in understanding the nature of cross-modal interactions and object formation – can’t wait!
What do you advise students interested in neuroscience as a profession, in terms of prerequisites, training, etc.?
Must admit I’m not the best person to ask for the most direct path to neuroscience. I would say it found me rather than the reverse. My studies have generally been driven by figuring out what tools and skills were needed to answer the questions I wished to ask. Having spent the majority of my undergraduate degree in a music conservatory (also working in labs and taking interesting science electives) neuroscience was where I ended up when I found the right level of reduction in my questions and answers that were satisfying for me. Absolutely -- I have at many times (and am sure will in many more) found myself in a situation of having to get up to speed in a new method or background simply because I hadn’t been exposed to it before – that’s science/life and just part of the day – also one of my favorite parts of being in a field that is constantly growing. In short – my advice is to choose what you are passionate about. No doubt, good programming skills (CS classes), a strong biological foundation, cognitive science, philosophy to keep you questioning yourself, and a good foundation of math/statistics/experimental design are very useful. However, probably one of the most useful things you can do as an undergraduate is seek out one or more labs to work in during your time in school. You will learn many skills, gain close relationships with professors and postdocs that will be valuable for both mentoring and recommendation letters, potentially have publications, and figure out what you do and don’t enjoy doing -- it is always hard to trump the drive, thought, and creativity that are likely to come from choosing your passions.
Thank you for sharing your expertise in this discussion! One question clearly leads to many more in the wide world of auditory neuroscience. Cogito is excited to showcase an exploration of this topic.