46 healthy human subjects (23 men, 23 women, age range: years)

Transcription

1 Supplementary Methods Tracking General Task 46 healthy human subjects (23 men, 23 women, age range: years) participated in this study. All subjects gave informed consent to procedures approved by the University of California Berkeley Committee for the Protection of Human Subjects. A consistent scent trail was generated across trials by embedding in the grass a twine line previously soaked in a 2% solution of odorant diluted into water. The concentration of odorant in the air above the ground was supra-threshold, but could not be characterized precisely given the turbulent nature of odorant plumes in natural environments. The scent path was set to form two straight trail segments joined by a ~45 degree angle. Subjects wore an opaque mask to block visual input, earmuffs to greatly reduce auditory input, and thick kneepads, elbow pads and work gloves to eliminate small-scale somatosensory input. In other words, the nose remained the only unobscured source of distal sensory input. Real-time nasal airflow was measured with a pressure transducer and recorded to a wireless data logger, both worn by the subject in a small backpack. The subject s position in the field was tracked using an aerial digital video recorder. Each trial started from about 3 meters away from the trail, with the subject situated in the general direction of the trail. The subject, crawling with nose to the ground, had to first locate the trail, and then follow it. The trial was terminated once the subject reached the end of the trail, or at 10 minutes, whichever came first. No feedback was provided during the task. 1

2 Odorant Concentration and Trigeminality The odorant concentration used to soak the twine in this experiment was 2% by volume, diluted into water. We tested 10 healthy subjects using the ML-PEST method 1, and found that their mean detection threshold for this odorant was 7.57 * 10 5 % +/ * 10 5, or roughly 4 log steps lower than the stimulus used. In addition we tested 2 anosmic subjects, to check if the concentration of odorant used activated the trigeminal pathway. Each subject scored in the anosmic range on the UPSIT, at 10 and 12 of 40 correct. Each subject performed a 10 trial 2 alternative forced choice task with a length of twine soaked in 2% perfume oil and a length of twine soaked in pure water. Neither subject was able to distinguish the twine soaked in chocolate from the control soaked in water (2/10 and 4/10 correct, both P > 0.05). Both subjects were able to distinguish propionic acid, a trigeminal odorant, from water (10/10 and 8/10 correct, both P < 0.05). Training During training, in order to ensure that subjects were improving their tracking ability, rather that just forming a motor memory of the path, the approach position was varied from run to run and the relative lengths of the two line segments was varied from day to day. The total track distance and mid-track angle were held constant across runs in order to maintain an equivalent task difficulty across training. Nasal Clip Control To ask whether this ability may have depended on some unintended non-olfactory cue in our experimental arrangement, in Experiment Two we repeated the task with 11 subjects wearing a plastic nasal clip that blocked sniffing. Subjects were again instructed to find and follow the track, this time using any means they could come up with to aid their search. None of the subjects was able to find or follow the trail under these conditions (this 2

3 is significantly different from Experiment 1: binomial sign test, P < ), suggesting that non-olfactory cues were not responsible for the performance of subjects in Experiment One. One-Nostril Tracking Fourteen subjects performed the same scent-tracking task as in Experiment One, once with one nostril taped closed, and once with both nostrils open. The order of conditions was counter-balanced across subjects. Nine of 25 trials (36%) resulted in successful tracking, a significantly lower rate than the 66% success rate observed with both nostrils open (binomial sign test: P < 0.003). Furthermore, subjects were slower on 8 of the 9 trials when using one nostril compared with two (on average a 26% reduction in speed, binomial sign test P < 0.02). Although this result suggested that dual-nostril information enhanced scenttracking, two alternative explanations may equally explain the poorer performance with one nostril as compared to two. First, it was possible that subjects performed less well simply because they had access to a weaker overall odor signal when sniffing with only one nostril rather than two. Second, it was also possible that bilateral neural processing is advantageous regardless of any spatial offset in the input. Blocking one side of the olfactory system may have restricted such bilateral neural processing. Therefore we went on to conduct Experiment Four in the main manuscript to address these concerns. Sniffing Power Spectra Dynamic power spectra were calculated for each trial using a short-time Fourier transform (window size = 12.8 s, overlap = 6.4 s, see Supplemental Figure 1a for an example dynamic power spectrum of sniffing behavior during a tracking run). These power spectra were used to calculate overall mean sniffing frequencies. 3

4 Across subjects the mean sniffing frequency on the first day of tests was 0.34 Hz, or 1 sniff every 2.96 seconds (n = 13, see Fig 2c) and the mean tracking velocity was 0.03 ms 1 (n = 13, see Fig 2c). On the final day of training the mean sniffing frequency was 0.66 Hz, or one sniff every 1.5 seconds, and the mean tracking velocity was 0.06 ms 1. Particle Image Velocimetry In Experiment Three, a 23-year-old healthy human male volunteer (JAV) sniffed in while the direction and velocity of neutrally buoyant particles in the air were measured using particle image velocimetry 2. The subject stood, with his chin resting on a horizontal rod for stability. A rectangular box (25.0 x 22.0 x 7.5 cm) was placed 3 cm beneath the subject s chin, a vertical distance of 12 cm from his nose. A 1.0 x 25.0 cm slit was located on the top face of the box. Water-based theatrical fog, consisting of particles of approximately 5 m in diameter, was injected into the box. Thus, this box served as a plenum or stilling chamber from which the subject inspired particle-laden air. For uniform seeding, a loosely woven cotton gauze fabric covered the slit on the top face of the box. A 0.5-mm thick vertical laser light sheet from a 200 mj double-pulse ND:Yag laser illuminated the seeded inspired air stream of the human subject in a coronal plane that intersected each nostril, more rostrally than caudally (see Fig. 3a). A CCD digital camera was used to capture 6 ensembles of 10 images pairs per second for 5 seconds, while the subject sniffed at approximately 0.2 Hz (a 5 second inspiration, followed by a short expiration). Of these 300 images pairs, 113 were selected for analysis based on the requirement of adequate seeding in the inhaled flows of both nostrils. Only velocities on inspiration were measured, due to the fact that only the inhaled air could be effectively seeded with particles The entire dataset was obtained during a 10-minute period to 4

5 minimize the effect of the nasal cycle 3. As shown previously 4, given care in particle seeding, theatrical fog particles are neutrally buoyant in the surrounding air. 5

6 Supp Fig. 1 An example dynamic power spectra and tracking path. (a) Dynamic power spectrum shows the power spectra as a function of time. Power spectra were calculated via a short time Fourier transform with a time window of 12.8s and an overlap of 6.4s. (b) A subject tracking path, shown in solid black line, with colored markers overlaid representing the instantaneous velocity at each point (ms 1 ). The subject encounters the scent trail at 35 seconds and encounters the sharp turn at 80 seconds. Inset is an enlarged view of a section of the path. The dashed line represents the underlying scent trail. 6

7 References: 1. Harvey, L.O., Jr. Efficient estimation of sensory thresholds with ML-PEST. Spat Vis 11, (1997). 2. Raffel, M., Willert, C., Kompenhans, J. Particle Image Velocimetry (Springer, Berlin, 1998). 3. Hasegawa, M. & Kern, E.B. The human nasal cycle. Mayo Clin Proc 52, (1977). 4. Gibbons, B. The Intimate Sense of Smell / National Geographic Smell Survey. National Geographic 170, (1986). 7