Sparse odor representation and olfactory learning

Abstract

Sensory systems create neural representations of environmental stimuli and these representations can be associated with other stimuli through learning. Are spike patterns the neural representations that get directly associated with reinforcement during conditioning? In the moth Manduca sexta, we found that odor presentations that support associative conditioning elicited only one or two spikes on the odor's onset (and sometimes offset) in each of a small fraction of Kenyon cells. Using associative conditioning procedures that effectively induced learning and varying the timing of reinforcement relative to spiking in Kenyon cells, we found that odor-elicited spiking in these cells ended well before the reinforcement was delivered. Furthermore, increasing the temporal overlap between spiking in Kenyon cells and reinforcement presentation actually reduced the efficacy of learning. Thus, spikes in Kenyon cells do not constitute the odor representation that coincides with reinforcement, and Hebbian spike timing-dependent plasticity in Kenyon cells alone cannot underlie this learning.

Figure 1

Projection neurons respond reliably to odors, and different odors evoke different temporally structured patterns of activity. (a) Examples of intracellular recordings of projection neurons (PN) responding to 4-s odor pulses (stimulus duration indicated by horizontal bars). Top, intracellular record of 1 trial. Bottom, rasters showing spikes from multiple trials. In PN1, 1% linalool induced brief inhibition followed by sustained spiking that outlasted the stimulus and a prolonged period of inhibition at the offset. In PN2, 1% cyclohexanone evoked only brief excitation. PN3, PN10 and PN12 showed distinct patterns to the same odor (100% hexanol). PN3 and PN10 showed excitatory off responses as well. Vertical scale bars represent 40 mV. (b) Peri-stimulus time histograms (PSTHs) showed reliable odor responses in projection neurons to 4-s odor pulses. These firing patterns contained information about odors (see Supplementary Fig. 2). Spikes were binned (10 ms) and bins with at least one spike are indicated by a black dot. One row represents one trial, and 62 projection neuron–odor combinations, each separated by a horizontal black bar, are shown. All projection neurons (except PN14) were tested with more than one odor.

Figure 2

Odor-elicited spiking in Kenyon cells is brief and sparse. (a) Examples of Kenyon cells (KC) responding to a panel of 21 odors. KC50a, KC41a and KC21a responded very sparsely, with either spikes at odor onset or offset. KC3a responded to a broader set of odors. KC1a was the most responsive cell in our set and fired reliably at different points in time for different odors. Ten trials were carried out for each odor. Rasters indicate spike times and the gray blocks indicate odor stimulation (4 s). See Methods for odors. (b) Spiking and subthreshold depolarization in Kenyon cells occurred mainly on odor onset and offset. The top trace indicates the intracellular voltage record and the dark horizontal line indicates odor delivery (4 s). The subsequent lines indicate the number of trials (one line per trial), and the rasters indicate spikes. Insets, enlarged membrane potential, averaged over first five trials, for on and off responses (times indicated as horizontal lines below rasters). (c) Histogram of Kenyon-cell firing probability (117 Kenyon cells, 10 trials each of 21 odors). The top brackets indicate the percentage of spikes during onset, middle and offset periods. The bottom brackets indicate the analysis bins used in subsequent panels. (d,e) Responses of Kenyon cells to odors were sparse. (e) Odor responses usually consisted of a single spike. Frequency distributions of odor-evoked spikes per trial measured over the full analysis bin are shown. (f) Different Kenyon cell ensembles were usually active during on, middle and off responses (ON, MD and OFF, respectively). MD&ON, overlap in spiking between middle and on responses; OFF&ON, overlap between off and on responses; OFF&MD, overlap with off and middle responses.

Figure 3

Kenyon cells responded only to the onset of brief odor pulses and to the onset and offset of long pulses. (a) Kenyon cell responses varied with odor pulse duration. Pulses at least 4 s long were most likely to induce odor-specific off responses. Briefer odor pulses generally elicited weaker or no off responses (see also b). Examples shown were selected for prominent off responses. Trials are shown from top to bottom (20 trials of 4-s odor pulses, then 6 shorter pulses, 3 trials each). (b) Off response probability increased with stimulus duration. Multiunit recordings of Kenyon cells (including 117 sorted cells from 16 animals, see Methods) responding to odor pulses of different durations (gray bars, tested in a randomized order, ten trials each). The histogram (bin size, 1 ms) combines the responses to five odors. (c) Long 4-s (black) and 18-s (gray) odor pulses evoked comparable onset and offset responses (arrows indicate the corresponding off responses). Multiunit recordings of Kenyon cells averaged across the four odors shown in d and across multiple trials are shown. (d) Examples of Kenyon cells responding to the offset of 18-s odor pulses. See Methods for the odor labels in a and d.

Figure 4

Greater temporal overlap between odor-elicited spiking in Kenyon cells and reinforcement delivery did not lead to more learning. (a) Diagrams illustrate PER conditioning procedures used to vary temporal overlap between spiking in Kenyon cells and sucrose delivery. Black traces represent time course of Kenyon cell spike response probability and gray boxes indicate analysis time windows used to compute conditioned stimulus (CS, odor)-elicited Kenyon cell spike probability concurrent with the unconditioned stimulus (US, sucrose) presentation shown in b (right ordinate). Conditioned stimulus was always paired with the unconditioned stimulus five times with 5 min between trials. The unconditioned stimulus duration was always 3 s. Short-term memory was tested 5 min after training by presenting the conditioned stimulus without the unconditioned stimulus. (b,c) Bar graphs in b and c show the PER probability for short-term memory tests. Asterisks indicate significant difference (P < 0.05, Fisher’s exact test with Bonferroni correction). More Kenyon cell spikes in the unconditioned stimulus period did not result in better learning. Open circles indicate normalized numbers of Kenyon cell spikes during the unconditioned stimulus presentation period (see Fig. 3b); spike counts were normalized with respect to the maximum elicited during the on response procedure (0.25-s ISI). (c) Reinforcement provided following off response spiking in Kenyon cells does not support learning. (d) The most effective conditioning occurred when the unconditioned stimulus followed the burst of onset spiking in Kenyon cells by a delay of several seconds. A delay of 20 s elicited no learning. The graph shows PER probability during the short-term memory test for different on response procedure groups.