Speed and accuracy of olfactory discrimination in the rat

Supplementary Fig. 1. (pdf 2M)
Optical imaging of activity evoked by odorants used in the behavioral study. Representative images obtained using aliphatic alcohols (hexanol, heptanol, S(+)-octanol and R(-)-2-octanol) and acids (caproic acid and butyric acid). Each image shows a thresholded, pseudo-colored response image superimposed on an image of the vasculature. Note that odorants having the same functional group activated highly overlapping glomeruli. The color scale at the right indicates the mapping signal (relative range in reflectance, see Supplementary Methods). P, posterior; L, lateral.

Supplementary Fig. 2. (pdf 178K)
Rapid performance holds in different experimental conditions. (a, b) Rapid performance is insensitive to interleaving of eight stimuli. Four rats were tested in a series of experiments using single complementary pairs of mixture ratios (80/20 and 20/80, etc.) in individual sessions using the odor pair, caproic acid versus hexanol (two sessions per condition). Odor sampling times obtained when only a single pair of mixture ratios was tested in a session (red) were similar to those obtained in the standard interleaved condition (black). (c, d) Rapid performance is insensitive to odor concentration. In order to test whether odor concentration would affect sampling strategy or accuracy, four rats were tested using S(+)- and R(-)-2-octanols at 100-fold lower concentration (mineral oil dilution). Performance accuracy and odor sampling times were similar between standard (black) and low concentration (red) conditions. (e, f) Rapid performance is insensitive to foreperiod. It has been observed that the variance of reaction times can be affected by the foreperiod (Roitman, J. D. & Shadlen, M. N. J. Neurosci. 22, 9475-89, 2002), the time between when the subject initiates the trial and the onset of the stimulus. In standard conditions, a uniform random delay of 0-100 ms between the detection of the nose poke and opening of the odor valve was used to prevent anticipation of odor onset. Three additional rats were tested using longer, exponentially distributed odor onset delays (i.e. a flat hazard function (Luce, R. D. Response Times: Their Role in Inferring Elementary Mental Organization 1986) with mean of 300 ms) on caproic acid versus hexanol. Odor sampling times were similar in standard (black) and longer foreperiod (red) conditions.

Supplementary Fig. 3. (pdf 163K)
Design of custom olfactometer. An olfactometer was constructed using small diameter (1/32" inner diameter) Teflon (PTFE) tubing and compression fittings to minimize dead space and delay times. (i) Flow rates of two air streams were independently controlled by mass flow controllers (range 2-100 ml/min) (100). A carrier air stream was controlled by an third flow controller (range: 20-1,000 ml/min) (1000) to produce 10:1 or greater dilution at a total flow rate of 1,000 ml/min. By mixing odorized air streams with defined flow rates, different mixture ratios were achieved. (ii) Two-way micro-solenoid valves controlled the timing of odor delivery. (iii) Saturated odor vapor was produced by flowing air across syringe filters loaded with liquid odorants (A, B). To maintain constant flow rates, valves for blank filters were actuated when odor delivery were closed. (iv) Odor and carrier streams were mixed at a manifold downstream of all valves immediately before the odor sampling port. The manifold was constructed out of chemically-inert polyetheretherketone (PEEK) material.

Supplementary Fig. 4. (pdf 1.8M)
Intrinsic signal imaging and analysis. (a) Intrinsic signals evoked by aliphatic acids and alcohols (as indicated above images). The top left panel shows a negative control image (pure air). All images were taken from the same rat. Signal intensity scale is indicated on the right, where negative values indicate darkening (activation). (b) Positions of identified glomeruli (yellow circles) superimposed on the vasculature image. P, posterior; L, lateral. (c) Summary of patterns of intrinsic signals in identified glomeruli. Black circles indicate average signal intensity and red circles indicate standard deviation (SD). Scales for average signal and SD are the same and shown at the upper right. Note that signal intensities have been inverted to positive values. Glomeruli were numbered by the position from anterior (A) to posterior (P) in (b).

Supplementary Fig. 5. (pdf 168K)
Cluster analysis of glomerular activation patterns calculated using different similarity metrics. The glomerular activity pattern evoked by each odorant (rows in Supplementary Fig. 4c) was treated as a vector and cluster analysis was performed using normalized and non-normalized methods for calculating vector distance (dissimilarity). (a) Clustering calculated using a normalized distance metric, 1 - cos(), where is the angle between the two vectors (b) Clustering calculated using a Euclidian distance metric. Note that the two methods produced similar patterns with the exception of a minor difference in the alcohol sub-cluster.

Supplementary Methods (pdf 24K)

Supplementary Video. (mov 2.8M)
Rat performing the odor mixture discrimination task. Binary mixtures of stereoisomers, S(+)-2-octanol (odor A) and R(-)-2-octanol (odor B), were delivered from the center port. The correct choice (the dominant component in the mixture) in each trial is indicated by a letter at the center which appears from the beginning of each trial to the beginning of nose poke. The rat sampled the odor at the central odor port and made a choice poke into left or right choice port (indicated by A and B, respectively). An interval of 4 s was imposed between choice poke and the beginning of the next trial. Note that the third trial choice was incorrect.

Top