Imagine bringing your groceries home and tucking them into the refrigerator. You’ll probably organize the items by categories: lemons and oranges into the fruit drawer, carrots and cauliflower into the vegetable drawer. Categorization is essential, allowing us to interact with the world in the most efficient way possible. If the differences between objects are not relevant to the task at hand, the brain will group objects together based on their shared properties and develop mental representations of the “categories”. Importantly, we are still aware of the distinctions between objects within the same category.

Categories of odor (for example, minty or fruity) are represented in a part of the brain called the olfactory (or piriform) cortex, which receives information from odor cues as well as “top-down” information from other areas of the brain. But how do these top-down pathways influence odor categorization?

Bao et al. asked how the brain solves the problem of categorizing odors. For the experiments, human volunteers smelled six familiar odors belonging to three different categories while their brain activity was monitored using a magnetic resonance imaging (fMRI) scanner. Then, half of the participants were given a drug called baclofen that prevents top-down inputs, but not odor cues, from reaching the piriform cortex, while the rest received a placebo. After five days of taking the medication, all of the volunteers had another session of fMRI where they had to categorize the same odors as before.

The experiments show that when comparing the fMRI scans before and after the drug treatment, the representations of odors belonging to the same category became more distinct in the piriform cortex in the placebo group. Put differently, as the volunteers were repeatedly exposed to odors of well-known categories, they became better at discriminating individual odors within the same category. However, these changes were disrupted in the group of volunteers that took baclofen.

Bao et al.’s findings indicate that this “practice makes perfect” approach to recognizing odors relies on top-down inputs into the piriform cortex. In future work it will be important to study the roles of these inputs in learning new categories of odors, and to investigate whether the mechanisms identified here apply to other sensory information and to more abstract knowledge.

Object categorization is an adaptive function of the brain, allowing organisms to sort information from the external world into behaviorally relevant classes. Importantly, sensory systems must generalize across different objects sharing similar features, but at the same time maintain the specificity of individual objects and categories (Roach, 1978; Riesenhuber and Poggio, 2000). Mechanisms of pattern recognition have been proposed to underlie the neural basis of object categorization, which requires a balance between generalizing inputs across a certain range of variations (known as pattern completion) and discriminating between distinct inputs (known as pattern separation) (Riesenhuber and Poggio, 2000; Haberly, 2001; Wilson and Sullivan, 2011; Chapuis and Wilson, 2012). Such computations can be achieved by associating sensory inputs with internal templates that are established through a lifetime of experience and encoded into memory (Bar, 2007).

Most neuroscientific research on pattern recognition has concentrated on the visual system, where associative areas in the visual ventral stream and the CA3 region of the hippocampus have been shown to support processes of object categorization (Riesenhuber and Poggio, 2000; Yassa and Stark, 2011; Haxby et al., 2001). In the olfactory system, information in a whiff of scented air is transformed into distributed patterns of neural activity in the piriform cortex, with both animal and human studies demonstrating that different odor objects evoke distinguishable ensemble activity patterns without spatial topography (Wilson and Sullivan, 2011; Gottfried, 2010; Bekkers and Suzuki, 2013; Stettler and Axel, 2009; Howard et al., 2009). Recent work has revealed that fMRI multivariate patterns in posterior piriform cortex (PPC) encode not only odor identity, but also category information (e.g., minty or woody), whereby odor patterns belonging to the same category are more similar (more overlapping) than those across different categories (Howard et al., 2009). Despite these insights, the mechanisms by which olfactory inputs are organized into categorical percepts through their associations with olfactory cortical areas are poorly understood.

The neural architecture of the piriform cortex makes it an attractive model for investigating mechanisms of odor object recognition. As the largest subregion of primary olfactory cortex, the piriform cortex receives afferent (bottom-up) inputs from the olfactory bulb through the lateral olfactory tract, and extensive associative (top-down) inputs from higher-order association areas such as orbitofrontal cortex (OFC), amygdala, and entorhinal cortex (Carmichael et al., 1994; Johnson et al., 2000; Haberly and Price, 1978; Insausti et al., 1987; Insausti et al., 2002). This convergence of bottom-up and top-down projections, along with the presence of dense recurrent collaterals, is thought to support olfactory pattern recognition and associative learning (Haberly, 2001; Haberly and Bower, 1989; Wilson, 2009). For example, when confronted with highly overlapping odor mixtures, rats can learn to discriminate or ignore detectable differences between these mixtures, with piriform activity patterns exhibiting either separation (enhanced discrimination) or completion (enhanced generalization), respectively (Chapuis and Wilson, 2012). Evidence from humans has also pointed to PPC as a substrate for odor discrimination (Li et al., 2008) and categorization (Howard et al., 2009). Together these findings suggest that piriform cortex is capable of modulating pattern representations along a discrimination-generalization spectrum in order to encode behaviorally adaptive meaning through perceptual experience.

While theoretical modelling and empirical evidence propose that piriform associative connections are essential for odor recognition (Haberly, 2001), few studies have explicitly investigated the relative contributions of afferent inputs versus associative networks in supporting odor categorization. In a previous fMRI study, human subjects were deprived of afferent sensory input for one week, resulting in a reduction of odor-evoked mean activity in PPC, without alteration of pattern-based piriform representations of odor categories (Wu et al., 2012). Here we address the inverse question, namely, how attenuation of piriform associative connections influences odor category coding in primary sensory regions and higher-order cortical areas.

To this end, we took advantage of the GABA(B) receptor agonist, baclofen, to modify the relative balance between afferent and associative inputs within piriform cortex. Baclofen selectively suppresses synaptic transmission of association fibers into piriform cortex, but leaves afferent inputs from the olfactory bulb unaffected (Tang and Hasselmo, 1994). In vivo local application of baclofen in the piriform cortex of anesthetized rats modified the strength of odor-evoked responses of pyramidal neurons, by blocking broadly-tuned neurons and increasing odor-selective responses (Poo and Isaacson, 2011). In behaving animals, injection of baclofen into the piriform cortex following an olfactory fear conditioning session resulted in fear memory generalization, indicating that piriform associative connections are essential for consolidation of stimulus-specific memories (Barnes and Wilson, 2014).

Inspired by these animal studies, we conducted a double-blind, placebo-controlled drug study in human subjects to examine fMRI ensemble representations of familiar odor categories before and after treatment with baclofen. Given that odor object codes take the form of distributed ensemble patterns, we used multivariate fMRI analyses to characterize baclofen effects in olfactory areas found to represent categorical information. The placebo group served as a control to account for session-effect confounds between pre- and post-drug phases of the study. As such, we examined the effects of baclofen by comparing pre-to-post changes relative to those observed in placebo subjects (i.e., group-by-session interaction). We predicted that baclofen would disrupt associative connections, leading to perceptual and neural reorganization of odor categories in piriform cortex and in olfactory downstream areas including OFC, amygdala, entorhinal cortex, and hippocampus.

The experiment spanned 5 days (Figure 1a). On day 1, subjects underwent pre-drug cognitive and psychophysical testing and fMRI scanning (Figure 1c). They were subsequently administered either placebo (n = 18) or increasing doses of baclofen (n = 14) for 5 consecutive days, in a double-blind design. This 5-day schedule was adopted to reach a target dose of 50-mg baclofen while minimizing the occurrence of side effects (Terrier et al., 2011). After taking the final dose on day 5, subjects underwent the same testing and fMRI scanning procedures as in the pre-drug session. During scanning, subjects completed an olfactory categorization task, as well as a control visual categorization task to establish the sensory specificity of the imaging findings.

Figure 1

Download asset Open asset

General cognition and olfactory perception

We first established that baclofen did not generally compromise cognitive or perceptual performance. Specifically, we found no significant differences between baclofen and placebo groups on neuropsychological assessments of basic cognition, short-term memory, visual attention, or task switching (Table 1). We also collected subjective reports of sleepiness using the Stanford Sleepiness Scale (SSS) during test sessions, given that the most common adverse reaction to baclofen medication is transient drowsiness (RxList The internet Drug Index, 2007). Baclofen subjects reported feeling sleepier after taking the drug (Figure 2), though reaction times during the fMRI categorization task did not differ from placebo subjects (Table 1). Finally, we examined whether baclofen altered general odor perception. Placebo and baclofen groups did not differ on olfactory measures of detection threshold, identification, fine odor discrimination (Figure 3d), or intensity and pleasantness ratings (for stimuli used in the main fMRI experiment) (Table 1), thereby reducing the possibility that baclofen-induced changes in odor perception could have influenced the imaging results.

Table 1

Task	Placebo (n = 18)		Baclofen (n = 14)		P value of group × session interaction
	Pre	Post	Pre	Post
MMSE	29.89 ± 0.11	29.94 ± 0.06	29.93 ± 0.07	30.00 ± 0	0.86
Digit span (forward)	7.22 ± 0.22	7.72 ± 0.14	7.14 ± 0.31	7.43 ± 0.20	0.55
Digit span (backward)	6.00 ± 0.20	6.00 ± 0.29	5.57 ± 0.31	5.71 ± 0.27	0.69
Trail making test B (s)	48.32 ± 2.70	39.05 ± 2.01	57.62 ± 5.42	47.11 ± 5.67	0.85
Stanford sleepiness scale	2.44 ± 0.17	2.22 ± 0.21	1.93 ± 0.20	2.50 ± 0.33	0.041*
Sniffin’ Sticks (odor detection threshold)	7.08 ± 0.84	9.65 ± 1.06	7.82 ± 0.95	8.57 ± 1.01	0.22
UPSIT (odor identification)	36.28 ± 0.61	36.00 ± 0.56	34.57 ± 0.49	33.79 ± 0.63	0.55
α- vs. β-pinene triangle test (fine odor discrimination)	0.66 ± 0.05	0.72 ± 0.05	0.72 ± 0.05	0.73 ± 0.07	0.45
Odor intensity ratings	4.00 ± 0.32	4.13 ± 0.31	3.05 ± 0.18	2.91 ± 0.28	0.39
Odor pleasantness ratings	5.43 ± 0.16	5.63 ± 0.16	5.64 ± 0.13	5.63 ± 0.16	0.12
Odor category descriptor ratings (within – across)	7.47 ± 0.44	7.49 ± 0.36	7.44 ± 0.46	7.78 ± 0.34	0.59
Odor pairwise similarity ratings (within – across)	4.16 ± 0.60	5.14 ± 0.55	3.93 ± 0.32	4.53 ± 0.44	0.59
Odor categorization catch trial accuracy	0.87 ± 0.04	0.89 ± 0.03	0.81 ± 0.04	0.81 ± 0.04	0.80
Odor categorization catch trial RT (s)	3.29 ± 0.23	2.85 ± 0.15	3.89 ± 0.38	3.48 ± 0.34	0.93
Visual categorization catch trial accuracy	0.97 ± 0.01 (n = 14)	0.99 ± 0.004	0.97 ± 0.01 (n = 11)	0.96 ± 0.01	0.21
Visual categorization catch trial RT (s)	0.42 ± 0.02	0.40 ± 0.03	0.44 ± 0.04	0.52 ± 0.06	0.25

1. Data are shown for cognitive and olfactory tests, as well as for behavioral performance in fMRI experiments from placebo and baclofen groups in pre- and post-drug sessions. Scores are presented as mean ± s.e.m. P values reported are for the interaction effects between group and session, based on a 2-way ANOVA, with one between-group ‘drug’ factor (placebo/baclofen) and one within-subject ‘session’ factor (pre/post). *P < 0.05.

Figure 2

Download asset Open asset

Figure 3

Download asset Open asset

Category-specific ensemble codes in PPC, OFC, amygdala and pHIP

During the fMRI odor categorization task, the six odors were delivered in a pseudorandom order, and subjects were cued to sniff upon odor delivery. They were asked to pay attention to the quality of the odors throughout the task, and make category judgments during catch trials.

As olfactory information takes the form of distributed patterns of fMRI activity in the human brain (Howard et al., 2009; Wu et al., 2012), multivariate pattern analyses are well-suited for examining the impact of baclofen on odor pattern recognition. We first used a support vector machine (SVM) classifier to identify brain areas where odor category information is represented, among several anatomically defined regions of interest (ROIs) including piriform cortex, higher-order areas that directly project to piriform (olfactory subregion of OFC, amygdala, entorhinal cortex), and hippocampus (Figure 4a). This analysis was conducted for all subjects in the pre-drug session, in order to constrain our investigation of baclofen-induced drug effects to those ROIs that had robust odor category coding at baseline (thus independent of drug administration). We trained the SVM classifier on patterns evoked by one pair of odors belonging to different categories (e.g., C1 vs. M1), and then tested the classifier on patterns evoked by the complementary pair of odors from the same categories (e.g., C2 vs. M2; Figure 4b). Importantly, because training and test sets were based on data evoked by different odor identities, significant above-chance decoding is only possible if fMRI patterns encode category information independent of the specific odor identities. Across all subjects in the pre-drug session, we found significant above-chance decoding accuracy in PPC (t₃₁ = 2.05, P = 0.024), OFC (t₃₁ = 1.96, P = 0.029), amygdala (t₃₁ = 3.17, P = 0.0017), and posterior hippocampus (pHIP, t₃₁ = 1.90, P = 0.034; Figure 4c). All subsequent analyses were constrained to these four regions where fMRI ensemble patterns encode odor category information.

Figure 4

Download asset Open asset

Baclofen interferes with within-category pattern separation in PPC

In order to characterize the continuous degree of pattern similarity between stimuli (Nili et al., 2014), we next used a linear correlation analysis (Haxby et al., 2001; Howard et al., 2009; Kriegeskorte et al., 2008), which provides a more direct assessment of pattern overlap. Specifically, to examine how baclofen alters the categorical organization of odors, we assembled vectors of ensemble pattern activity from all voxels within PPC, and measured the dissimilarity (correlation distances) of pattern vectors evoked by across-category odors (e.g., C1/M1) and within-category odors (e.g., C1/C2, Figure 5a). In order to control for within-session and between-session variations that could arise from training, familiarity, increasing boredom, drug effect, scanner drift or other potential artifact, we computed pattern distances between the same odor as baseline patterns, and then subtracted these from the within-category and across-category pattern distances. A two-way mixed-model ANOVA on same-odor pattern distances showed that there was no main effects of session or group, or group × session interaction (F_1,30 = 1.88, P = 0.18), indicating that the baseline patterns were consistent across time and group, and were not affected by the drug. We then tested a three-way analysis of variance (ANOVA), with two within-subject factors of session (pre/post) and category type (within-/across-category), and one between-subject factor of drug (placebo/baclofen). This yielded a significant session × category type × drug interaction effect (F_1,30 = 5.49, P = 0.026) in the absence of other main effects or two-way interactions (all P’s >0.15), and suggests that baclofen significantly affected the categorical structure of odor pattern representations in PPC.

Figure 5

Download asset Open asset

Based on inspection of the odor-evoked pattern changes in PPC (Figure 5b), it is evident that these changes were actually more prominent in the placebo group, and for the within-category condition. To assess these hypotheses, we examined drug-related categorization effects separately in each group. In placebo subjects the interaction of session × category type was significant in PPC (F_1,17 = 9.35, P = 0.0071; repeated-measures ANOVA), whereas no such interaction was identified in the baclofen group (F_1,13 = 0.62, P = 0.45). This effect was driven by a significant increase of the within-category odor distance in the placebo group (F_1,17 = 5.23, P = 0.035), but not in the baclofen group (F_1,13 = 2.61, P = 0.13), with a significant difference between groups (F_1,30 = 7.36, P = 0.011; mixed-model ANOVA, session × group interaction). On the other hand, across-category odor distances did not differ for either group (placebo, F_1,17 = 0.75, P = 0.40; baclofen, F_1,13 = 0.27, P = 0.61) or between groups (F_1,30 = 0.00014, P = 0.99, Figure 5b).

These results highlight a divergence in PPC pattern representations for odors belonging to the same category, but only in the placebo group. One implication is that repeated exposure to the odors (in absence of drug) induced pattern separation or differentiation, a process that appears to be blocked in the presence of baclofen. Interestingly, this conceptualization – greater pattern separation over time in the control subjects – is in close accordance with an earlier olfactory perceptual learning study from our lab, where prolonged passive exposure to one target odor increased its discriminability from categorically related odors (Li et al., 2006). Viewed in this context, it is reasonable to speculate that baclofen interferes with the natural emergence of olfactory pattern separation in PPC, possibly reflecting a disruption in consolidation mechanisms that normally underlie perceptual learning.

If pattern separation in PPC is critical for differentiating categorically related odors, it follows that subjects with greater disruption of PPC pattern separation (as a result of baclofen treatment) should exhibit greater olfactory perceptual deficits. This hypothesis was tested by regressing subject-wise measures of fine odor discrimination (Figure 3d) against the magnitude of baclofen-induced pattern changes in PPC. We found a significant correlation between perceptual performance change and the degree of odor-evoked pattern separation in PPC (ρ = 0.51, P = 0.031, one-tailed; Figure 5d). Thus, subjects with less within-category odor separation in PPC showed greater difficulty in discriminating between odors sharing semantic features.

It is worth considering that because baclofen produced significant sleepiness, the associated tiredness and sedation may have also been associated with a lack of attention on the hardest discriminations. To investigate this possibility, we tested the correlation between pre-post changes in sleep scale ratings and changes in fine odor discrimination performance across baclofen subjects. This relationship was not significant (Spearman ρ = -0.33, P = 0.25, n = 14). Likewise, there was no significant correlation between sleep scale changes and within-category pattern distance changes in PPC across baclofen subjects (Spearman ρ = 0.11, P = 0.70, n = 14). These data suggest that there was no direct evidence of a systematic link between sleepiness and odor discrimination at the behavioral or neural level.

Baclofen disrupts category coding in OFC and impedes within-category generalization in pHIP

Because olfactory categorical codes were also identified in OFC, amygdala, and pHIP in the pre-treatment session (Figure 4), we also investigated the effects of baclofen on categorical organization of odor ensemble patterns in these regions. Significant three-way interactions of session × category type × drug were found in OFC (F_1,30 = 4.48, P = 0.043) and pHIP (F_1,30 = 5.90, P = 0.021) without other main effects or two-way interactions. No significant interaction was observed in amygdala (F_1,30 = 0.047, P = 0.83; Figure 6c).

Figure 6

Download asset Open asset

Following the same approach used for PPC, we next asked how odor category organization in OFC changes in each group. We found that in OFC, the session × category interaction approached significance in the baclofen group (F_1,13 = 4.14, P = 0.063), driven by a significant decrease in across-category distances (F_1,13 = 5.93, P = 0.030; Figure 6a), and leading to a net effect of category disruption (that is, greater convergence of across-category patterns; e.g., C1 and M1 becoming more alike, Figure 5a). There was no categorical change in the placebo group (session × category interaction, F_1,17 = 0.17, P = 0.69).

By contrast, in pHIP, there was a significant session × category interaction in the placebo group (F_1,17 = 7.68, P = 0.013), here driven by a trend decrease in within-category distances (F_1,17 = 3.09, P = 0.097, Figure 6b), and giving rise to an enhanced categorical structure among odors (that is, greater convergence of within-category patterns; e.g., C1 and C2 becoming more alike, Figure 5a). Of note, this profile is opposite to that seen in PPC (Figure 5b). On the contrary, there was no categorical change in the baclofen group (session × category interaction, F_1,13 = 0.77, P = 0.40).

Effects of baclofen are specific to olfactory processing

The above findings indicate that baclofen had selective effects on odor category coding in PPC, OFC, and pHIP. However, because baclofen was administered systemically, it remains unclear whether the effects were specific to odor categorization, or merely altered semantic or conceptual processing independently of sensory modality. Therefore, in a parallel fMRI experiment, the same subjects performed a visual categorization task (Figure 7a), viewing six images belonging to three categories (chairs, teapots, and houses) and identifying the category on catch trials. There was no effect of baclofen on response accuracies and reaction times (Table 1).

Figure 7

Download asset Open asset

We then utilized the same multivariate analysis pipeline to explore the effects of baclofen on visual pattern recognition. First we used the SVM classifier to decode visual category information in the same ROIs as in the olfactory task. We also included two additional visual ROIs located in lateral occipital complex (LOC) and fusiform gyrus as defined by an independent functional localizer scan, and which are known to be involved in visual object recognition (Haxby et al., 2001; Cox and Savoy, 2003; Grill-Spector, 2003; Kriegeskorte et al., 2008). Across all subjects in the pre-drug session, category decoding accuracy was significantly above chance in LOC (t₃₀ = 2.33, P = 0.013), but not in fusiform cortex (t₃₀ = -1.20, P = 0.88) or in any of the olfactory ROIs (PPC: t₃₀ = 0.36, P = 0.72; OFC: t₃₀ = -1.36, P = 0.18; amygdala: t₃₀ = -2.47, P = 0.99; pHIP: t₃₀ = 0.078, P = 0.47; Figure 7b). Next, we performed an fMRI pattern correlation analysis to test the drug effect on visual category representations. A three-way session × category type × drug interaction was not significant in LOC (F_1,29 = 0.46, P = 0.50; Figure 7c), suggesting that baclofen did not alter coding of categories in the visual domain.

Finally, we compared the effect of baclofen on categorization between olfactory and visual tasks, and found that the impact of baclofen on category coding in PPC was specific to olfaction. A mixed four-way ANOVA (three within-subject factors of modality, category type, and session; one between-subject factor of group) revealed a significant interaction of modality × category × session × drug (F_1,29 = 4.41, P = 0.044). Thus, while baclofen blocked within-category separation in PPC, it did not alter the visual categorization compared to placebo (category × session × drug interaction: F_1,29 = 0.056, P = 0.81). These findings imply that the observed effect of baclofen in PPC was not due to generic changes in semantic processing, nor to non-specific changes in hemodynamic parameters, but instead was due to alterations in information coding in the presence of olfactory inputs.

In this study we investigated the role of piriform associative connections in the neural coding of odor categories. We used the GABA(B) receptor agonist, baclofen, to reduce associative input in the olfactory network while sparing afferent input from the periphery. This pharmacological manipulation, combined with multivariate pattern analysis, enabled us to examine how baclofen treatment alters fMRI pattern representations of odors within and across categories relative to placebo. We found that in PPC, baclofen (compared to placebo) interfered with the emergent pattern separation of odors belonging to the same categorical class. The magnitude of this effect correlated with difficulties in fine-odor discrimination at the perceptual level. In contrast, baclofen disrupted across-category separation in olfactory downstream region of OFC, and impeded within-category generalization in pHIP.

Interestingly, the baclofen effect observed in PPC was opposite to our original prediction that baclofen would simply weaken the boundaries between categories, leading to reduced pattern separation between citrus, mint, and wood odors. Instead, there was increased pattern separation for within-category odors over time, but specifically in the placebo group, likely reflecting perceptual training or stimulus-specific consolidation from the pre- to post-session. The significant group × session interaction implies that the natural process of within-category pattern separation in controls was disrupted in the presence of baclofen. For example, PPC pattern representations of the two citrus odors became more distinct under the control condition, but failed to diverge under baclofen. We speculate that piriform associative input normally supports the separation of patterns corresponding to unique identities of individual odors, especially those sharing perceptual features and associated with the same semantic labels. This mechanism would be compatible with prior work showing that perceptual learning enhances discriminability of within-category odor pairs, with concomitant fMRI changes in PPC as well as OFC (Li et al., 2006).

It is worth considering why baclofen had no effect on across-category odor separation in PPC. One plausible explanation is that piriform cortex has the capacity to enhance either pattern separation or completion, as a function of task demands (Chapuis and Wilson, 2012; Li et al., 2008; Shakhawat et al., 2014). Various rodent paradigms of olfactory associative learning have shown that the direction of piriform pattern changes can flexibly match the behavioral requirements for either odor discrimination (i.e., pattern separation) or odor generalization (i.e., pattern completion). In the current experiment, subjects were asked to perform an odor categorization task, in which differences across categories, but not within category, were emphasized. As such, our experimental design might have helped stabilize category-specific differences in PPC, even in the presence of baclofen, though at the expense of within-category odor separation. The fact that categorical representations of citrus, mint, and wood odors were already highly familiar to the subjects also could have created further stability against across-category pattern changes. Thus, it is fair to say that while we have no evidence to support baclofen-induced disruption of across-category codes, the nature of our task leaves open the possibility that in a different task where categorical distinctions were less relevant, baclofen might have a modulatory influence on across-category differences.

Another potential factor related to the experimental design that might complicate interpretation of the PPC results is a short-term order effect between trials. That is to say, pattern representations of the same odor could differ depending on whether the preceding odor belonged to the same category (category repetition, e.g., C1 preceded by C2) or a different category (non-repetition, e.g., C1 preceded by M1). This repetition factor could induce short-term ‘learning’ or adaptation effects that involve local synaptic interactions mediated by GABA(B) receptors (Brenowitz et al., 1998; Ziakopoulos et al., 2000), and thus could be susceptible to baclofen manipulation. Examination of our data indicates that across subjects and pre/post sessions, there were 39.3 ± 0.6 (mean ± SE) category repetition trials as opposed to 121.5 ± 0.9 category non-repetition trials. That the majority (~75%) of trials in our experiment belonged to non-repetition trials suggests that category repetitions would not have had a pronounced effect on the findings. In additional analyses (see Materials and methods), the proportion of category repetition versus non-repetition trials (per imaging run) was not associated with the strength of pattern categorization effects in PPC, and PPC patterns evoked by the same odor under repetition and non-repetition conditions were not significantly different. Thus, it is reasonable to conclude that the sequence order of the odors did not have an impact on categorical pattern coding in PPC, either at baseline or in the context of drug.

In contrast to PPC, fMRI patterns in olfactory downstream areas, including OFC and pHIP, showed deficient category coding in the baclofen group. Thus in OFC, the discrete categorical patterns for citrus, mint, and wood became less separated, in the presence of baclofen. In spite of these changes, there was no parallel impact on behavior. Indeed, baclofen had no perceptual effect on categorical discrimination, and we would argue that such a finding would have been unlikely, presumably due to high familiarity and discriminability of odor categories. However, to the extent that the existence of an odor category necessitates an association between an olfactory stimulus and semantic conceptual knowledge, these results are consistent with the recognized integrative role of OFC in guiding olfactory-based behavior. Both animal and human studies have demonstrated that OFC patterns can differentiate between odor objects and categories (Howard et al., 2009; Wu et al., 2012; Schoenbaum and Eichenbaum, 1995; Critchley and Rolls, 1996). Moreover, the OFC has been proposed to integrate taste and visual information associated with odor stimuli (Critchley and Rolls, 1996; Gottfried and Dolan, 2003), encode the reward value of odors (Howard and Gottfried, 2014), disambiguate mixtures of categorically dissimilar odors (Bowman et al., 2012), and represent olfactory lexical-semantic content (Olofsson et al., 2014). Viewed in this context, our results highlight the role of OFC in preserving the perceptual distinctions between different odor categories, likely through its associative access to multimodal and semantic information streams.

The demonstration of olfactory category coding in pHIP, and its vulnerability to baclofen, echoes hippocampal findings in the visual modality (Seger and Miller, 2010; Seger and Peterson, 2013; Kumaran and McClelland, 2012). For example, single-unit recordings from the hippocampus have identified neurons in both humans and monkeys that are able to categorize visual information (Kreiman et al., 2000; Hampson et al., 2004), and fMRI activity in human hippocampus is selectively increased when memory performance relies on perceptual generalization across stimuli (Preston et al., 2004; Shohamy and Wagner, 2008). Considered in this framework, the trend effect of decreased within-category separation in pHIP in placebo subjects may reflect the role of hippocampus to generalize, or to make inferences, across shared odor features, essentially bringing odors of the same category closer together, and creating more separation between different categories (Figure 5a). It is interesting to note that both piriform cortex and hippocampus have long been regarded as canonical models of autoassociative networks where pattern separation and pattern completion computations can be flexibly achieved (Yassa and Stark, 2011; Bekkers and Suzuki, 2013; Wilson, 2009; Leutgeb and Leutgeb, 2007; Hunsaker and Kesner, 2013; LaRocque et al., 2013; Eichenbaum et al., 2007). That the effect of baclofen was to impede within-category discrimination in PPC, while simultaneously impeding within-category generalization in pHIP, highlights a unique functional difference between these two anatomically homologous regions, and may help bring new mechanistic understanding of the contributions of piriform cortex, hippocampus, and piriform-hippocampal interactions to human olfactory processing and perception.

Our behavioral data indicate that the 50-mg baclofen dose did not impair general cognition or olfactory perceptual performance, suggesting that off-target effects of the drug were minimal, other than a modest effect on subjective sleepiness that did not interfere with online task accuracy or response times. While it is possible that the 50-mg dose may not have been potent enough to exert a physiological effect, the study medication schedule was similar to those used in other human studies that administered baclofen to induce reliable changes in brain activity or behavior (Terrier et al., 2011; Franklin et al., 2012; Young et al., 2014; Franklin et al., 2011). In our study, we did not find evidence of drug effect on odor categorization behavior, in spite of significant changes in fMRI pattern representations. There are at least three possibilities to account for this discrepancy. One possibility is that our behavioral tests were simply not sensitive enough to detect changes in perceptual performance. Across both placebo and baclofen groups, there was a general trend towards improved performance, likely reflecting effects of training and exposure. As such, any further subtle effect of baclofen on perception may not have emerged beyond these training effects per se. Related to this, even if the three-way forced-choice pinene triangle test was arguably the most ‘sensitive’ or difficult test of odor discrimination, this test might not have revealed a significant change if the perceptual learning effects had been confined to those odors presented repeatedly during the fMRI. A second possibility is that because all of the subjects were explicitly informed of the categorical features of the odor stimuli, there may have been an implicit tendency to anchor their perceptual responses to semantic categorical attributes. Moreover, even at baseline, all of the odors were easy to discriminate and highly familiar, and subjects were regularly called upon to make category judgments of the odors throughout the experiment. Thus, even though there was reorganization of categorical representations in PPC, these factors could have obscured our ability to observe perceptual plasticity across the set of odors. Finally, while changes in piriform odor representations might have induced parallel changes in perception, it is possible that other brain areas would be able to compensate for these perturbations, helping to stabilize olfactory perceptual performance. For example, in the placebo group, increased within-category pattern separation in PPC (Figure 5b) would be counteracted by decreased within-category pattern separation in pHIP (Figure 6b), resulting in no detectable change at the behavioral level.

One potential issue is that baclofen can also target GABA(B) receptors that have been identified in area CA1 of the hippocampus, influencing visual object recognition and memory (Lanthorn and Cotman, 1981; Ault and Nadler, 1982). Therefore, to establish that our findings were specific to the olfactory system, and to ensure that baclofen did not disrupt general semantic processing and object categorization, subjects also performed a visual categorization fMRI task in which they viewed pictures rather than smelled odors. This control study confirmed that our pharmacological manipulation induced both regional and modality specificity, thus ruling out possible confounds such as altered global attention, arousal, or hemodynamic reactivity. As an added way to minimize mere drug effects, we explicitly focused our imaging analyses on the interactions between group (baclofen/placebo), session (pre/post), and category level (within/across), effectively cancelling out any other session-related confounds.

An unavoidable limitation of this study was that baclofen was administered systemically. While our findings demonstrate regionally selective treatment effects in PPC, it is not possible to confirm that these changes were due to the direct action of baclofen solely at piriform cortex. There are at least three mechanisms by which baclofen could affect categorization in the olfactory network, none of which are mutually exclusive. First, baclofen might directly target the layer 1b synapses in piriform cortex where associative intracortical and extracortical inputs predominate. This would most closely mirror what has been tested using focal baclofen injections in animal models (Poo and Isaacson, 2011; Barnes and Wilson, 2014), and would underscore the idea that categorical odor representations rely on associative information processing within this layer of piriform cortex. Second, baclofen might target neurons in OFC, entorhinal cortex, and other associative brain areas that project onto piriform cortex. Given that the fMRI BOLD response is thought to reflect local dendritic processing and population activity (Logothetis and Wandell, 2004; Hipp and Siegel, 2015), our findings could reflect a distant action of baclofen on OFC (or other areas), which in turn alters distributed fMRI patterns measured in piriform cortex. Third, the changes seen in PPC could theoretically have arisen in the olfactory bulb, where GABA(B) receptors have also been described (Nickell et al., 1994; Palouzier-Paulignan et al., 2002; Okutani et al., 2003; Wachowiak et al., 2005; Aroniadou-Anderjaska et al., 2000; Isaacson and Vitten, 2003; Karpuk and Hayar, 2008). In this instance, one might have predicted a more profound olfactory perceptual deficit, including impairments of odor threshold, identification, and perceived intensity, though such a profile was not found in our study. Irrespective of the specific mechanism or mechanisms, these findings establish a critical role of the GABA(B) receptor in modulating categorical representations in PPC and OFC, with specificity for the olfactory modality.

In summary, our study provides a foundation for understanding the contribution of afferent and associative inputs to odor categorical perception in the human brain. Of note, this work forms a counterpoint to an earlier study from our lab in which subjects underwent a 7-day period of odor deprivation (Wu et al., 2012): by reducing olfactory afferent input, we were able to show that multivariate pattern representations of odor category were selectively altered in OFC, without any pattern-based changes observed in PPC. By comparison, in the current study, we were able to test the inverse manipulation, using baclofen to reduce olfactory associative input. In this instance, we again observed a disruption of odor categorization in OFC, but also an interference of session-related pattern changes in PPC and pHIP. The fact that within-category pattern changes in PPC were complementary to those in pHIP, in conjunction with the different course of across-category changes in OFC, underscores the idea that odor categorization is a dynamic process involving multiple stages of an extended olfactory network. We surmise that under normal conditions, the ability to refine discriminability of within-category odors in PPC through experience helps to improve perceptual acuity and decision making, and to prevent perceptual generalization from becoming maladaptive. With the interruption of associative input, in the setting of experimental baclofen or even perhaps as the consequence of a neurological disorder, within-category boundaries can become obscured, leading to perceptual over-generalization that can result in detrimental choices. As such, our findings may point toward an important mechanism by which associative networks regulate perceptual processing. Whether such mechanisms are restricted to the olfactory modality, or apply more widely across different sensory systems, remains to be determined.

1. 1. Aroniadou-Anderjaska V
   2. Zhou FM
   3. Priest CA
   4. Ennis M
   5. Shipley MT

   (2000)

   Tonic and synaptically evoked presynaptic inhibition of sensory input to the rat olfactory bulb via GABA(B) heteroreceptors

   Journal of Neurophysiology 84:1194–1203.

   • Google Scholar
2. 1. Ault B
   2. Nadler JV

   (1982)

   Baclofen selectively inhibits transmission at synapses made by axons of CA3 pyramidal cells in the hippocampal slice

   The Journal of Pharmacology and Experimental Therapeutics 223:291–297.

   • Google Scholar
3. 1. Bar M

   (2007) Corrigendum: The proactive brain: Using analogies and associations to generate predictions

   Trends in Cognitive Sciences 11:372.

   https://doi.org/10.1016/j.tics.2007.08.004

   • Google Scholar
4. 1. Barnes DC
   2. Wilson DA

   (2014) Slow-wave sleep-imposed replay modulates both strength and precision of memory

   The Journal of Neuroscience 34:5134–5142.

   https://doi.org/10.1523/JNEUROSCI.5274-13.2014

   • Google Scholar
5. 1. Bekkers JM
   2. Suzuki N

   (2013) Neurons and circuits for odor processing in the piriform cortex

   Trends in Neurosciences 36:429–438.

   https://doi.org/10.1016/j.tins.2013.04.005

   • Google Scholar
6. 1. Bowie CR
   2. Harvey PD

   (2006) Administration and interpretation of the trail making test

   Nature Protocols 1:2277–2281.

   https://doi.org/10.1038/nprot.2006.390

   • Google Scholar
7. 1. Bowman NE
   2. Kording KP
   3. Gottfried JA

   (2012) Temporal integration of olfactory perceptual evidence in human orbitofrontal cortex

   Neuron 75:916–927.

   https://doi.org/10.1016/j.neuron.2012.06.035

   • Google Scholar
8. 1. Brenowitz S
   2. David J
   3. Trussell L

   (1998) Enhancement of synaptic efficacy by presynaptic GABA(B) receptors

   Neuron 20:135–141.

   https://doi.org/10.1016/S0896-6273(00)80441-9

   • Google Scholar
9. 1. Carmichael ST
   2. Clugnet MC
   3. Price JL

   (1994) Central olfactory connections in the macaque monkey

   The Journal of Comparative Neurology 346:403–434.

   https://doi.org/10.1002/cne.903460306

   • Google Scholar
10. 1. Chang C-C
    2. Lin C-J

    (2011) LIBSVM

    ACM Transactions on Intelligent Systems and Technology 2:1–27.

    https://doi.org/10.1145/1961189.1961199

    • Google Scholar
11. 1. Chapuis J
    2. Wilson DA

    (2012) Bidirectional plasticity of cortical pattern recognition and behavioral sensory acuity

    Nature Neuroscience 15:155–161.

    https://doi.org/10.1038/nn.2966

    • Google Scholar
12. 1. Cox DD
    2. Savoy RL

    (2003) Functional magnetic resonance imaging (fmri) "brain reading": Detecting and classifying distributed patterns of fmri activity in human visual cortex

    NeuroImage 19:261–270.

    https://doi.org/10.1016/S1053-8119(03)00049-1

    • Google Scholar
13. 1. Critchley HD
    2. Rolls ET

    (1996)

    Olfactory neuronal responses in the primate orbitofrontal cortex: Analysis in an olfactory discrimination task

    Journal of Neurophysiology 75:1659–1672.

    • Google Scholar
14. 1. Doty RL
    2. Shaman P
    3. Kimmelman CP
    4. Dann MS

    (1984) University of pennsylvania smell identification test: A rapid quantitative olfactory function test for the clinic

    The Laryngoscope 94:176–178.

    https://doi.org/10.1288/00005537-198402000-00004

    • Google Scholar
15. 1. Eichenbaum H
    2. Yonelinas AP
    3. Ranganath C

    (2007) The medial temporal lobe and recognition memory

    Annual Review of Neuroscience 30:123–152.

    https://doi.org/10.1146/annurev.neuro.30.051606.094328

    • Google Scholar
16. 1. Folstein MF
    2. Folstein SE
    3. McHugh PR

    (1975) "Mini-mental state". A practical method for grading the cognitive state of patients for the clinician

    Journal of Psychiatric Research 12:189–198.

    https://doi.org/10.1016/0022-3956(75)90026-6

    • Google Scholar
17. 1. Franklin TR
    2. Shin J
    3. Jagannathan K
    4. Suh JJ
    5. Detre JA
    6. O'Brien CP
    7. Childress AR

    (2012) Acute baclofen diminishes resting baseline blood flow to limbic structures: A perfusion fmri study

    Drug and Alcohol Dependence 125:60–66.

    https://doi.org/10.1016/j.drugalcdep.2012.03.016

    • Google Scholar
18. 1. Franklin TR
    2. Wang Z
    3. Sciortino N
    4. Harper D
    5. Li Y
    6. Hakun J
    7. Kildea S
    8. Kampman K
    9. Ehrman R
    10. Detre JA
    11. O'Brien CP
    12. Childress AR

    (2011) Modulation of resting brain cerebral blood flow by the GABA B agonist, baclofen: A longitudinal perfusion fmri study

    Drug and Alcohol Dependence 117:176–183.

    https://doi.org/10.1016/j.drugalcdep.2011.01.015

    • Google Scholar
19. 1. Gottfried JA
    2. Dolan RJ

    (2003) The nose smells what the eye sees: Crossmodal visual facilitation of human olfactory perception

    Neuron 39:375–386.

    https://doi.org/10.1016/S0896-6273(03)00392-1

    • Google Scholar
20. 1. Gottfried JA
    2. Zald DH

    (2005) On the scent of human olfactory orbitofrontal cortex: Meta-analysis and comparison to non-human primates

    Brain Research. Brain Research Reviews 50:287–304.

    https://doi.org/10.1016/j.brainresrev.2005.08.004

    • Google Scholar
21. 1. Gottfried JA

    (2010) Central mechanisms of odour object perception

    Nature Reviews. Neuroscience 11:628–641.

    https://doi.org/10.1038/nrn2883

    • Google Scholar
22. 1. Grill-Spector K

    (2003) The neural basis of object perception

    Current Opinion in Neurobiology 13:159–166.

    https://doi.org/10.1016/S0959-4388(03)00040-0

    • Google Scholar
23. 1. Haberly LB
    2. Bower JM

    (1989) Olfactory cortex: Model circuit for study of associative memory?

    Trends in Neurosciences 12:258–264.

    https://doi.org/10.1016/0166-2236(89)90025-8

    • Google Scholar
24. 1. Haberly LB
    2. Price JL

    (1978) Association and commissural fiber systems of the olfactory cortex of the rat. II. systems originating in the olfactory peduncle

    The Journal of Comparative Neurology 181:781–807.

    https://doi.org/10.1002/cne.901810407

    • Google Scholar
25. 1. Haberly LB

    (2001) Parallel-distributed processing in olfactory cortex: New insights from morphological and physiological analysis of neuronal circuitry

    Chemical Senses 26:551–576.

    https://doi.org/10.1093/chemse/26.5.551

    • Google Scholar
26. 1. Hampson RE
    2. Pons TP
    3. Stanford TR
    4. Deadwyler SA

    (2004) Categorization in the monkey hippocampus: A possible mechanism for encoding information into memory

    Proceedings of the National Academy of Sciences of the United States of America 101:3184–3189.

    https://doi.org/10.1073/pnas.0400162101

    • Google Scholar
27. 1. Haxby JV
    2. Gobbini MI
    3. Furey ML
    4. Ishai A
    5. Schouten JL
    6. Pietrini P

    (2001) Distributed and overlapping representations of faces and objects in ventral temporal cortex

    Science 293:2425–2430.

    https://doi.org/10.1126/science.1063736

    • Google Scholar
28. 1. Hipp JF
    2. Siegel M

    (2015) BOLD fmri correlation reflects frequency-specific neuronal correlation

    Current Biology 25:1368–1374.

    https://doi.org/10.1016/j.cub.2015.03.049

    • Google Scholar
29. 1. Hoddes E
    2. Zarcone V
    3. Smythe H
    4. Phillips R
    5. Dement WC

    (1973) Quantification of sleepiness: A new approach

    Psychophysiology 10:431–436.

    https://doi.org/10.1111/j.1469-8986.1973.tb00801.x

    • Google Scholar
30. 1. Howard JD
    2. Gottfried JA

    (2014) Configural and elemental coding of natural odor mixture components in the human brain

    Neuron 84:857–869.

    https://doi.org/10.1016/j.neuron.2014.10.012

    • Google Scholar
31. 1. Howard JD
    2. Plailly J
    3. Grueschow M
    4. Haynes JD
    5. Gottfried JA

    (2009) Odor quality coding and categorization in human posterior piriform cortex

    Nature Neuroscience 12:932–938.

    https://doi.org/10.1038/nn.2324

    • Google Scholar
32. 1. Hummel T
    2. Sekinger B
    3. Wolf SR
    4. Pauli E
    5. Kobal G

    (1997) 'Sniffin' sticks': Olfactory performance assessed by the combined testing of odor identification, odor discrimination and olfactory threshold

    Chemical Senses 22:39–52.

    https://doi.org/10.1093/chemse/22.1.39

    • Google Scholar
33. 1. Hunsaker MR
    2. Kesner RP

    (2013) The operation of pattern separation and pattern completion processes associated with different attributes or domains of memory

    Neuroscience and Biobehavioral Reviews 37:36–58.

    https://doi.org/10.1016/j.neubiorev.2012.09.014

    • Google Scholar
34. 1. Insausti R
    2. Amaral DG
    3. Cowan WM

    (1987) The entorhinal cortex of the monkey: II. cortical afferents

    The Journal of Comparative Neurology 264:356–395.

    https://doi.org/10.1002/cne.902640306

    • Google Scholar
35. 1. Insausti R
    2. Juottonen K
    3. Soininen H
    4. Insausti AM
    5. Partanen K
    6. Vainio P
    7. Laakso MP
    8. Pitkänen A

    (1998)

    MR volumetric analysis of the human entorhinal, perirhinal, and temporopolar cortices

    AJNR. American Journal of Neuroradiology 19:659–71.

    • Google Scholar
36. 1. Insausti R
    2. Marcos P
    3. Arroyo-Jiménez MM
    4. Blaizot X
    5. Martínez-Marcos A

    (2002) Comparative aspects of the olfactory portion of the entorhinal cortex and its projection to the hippocampus in rodents, nonhuman primates, and the human brain

    Brain Research Bulletin 57:557–560.

    https://doi.org/10.1016/S0361-9230(01)00684-0

    • Google Scholar
37. 1. Isaacson JS
    2. Vitten H

    (2003)

    GABA(B) receptors inhibit dendrodendritic transmission in the rat olfactory bulb

    The Journal of Neuroscience 23:2032–2039.

    • Google Scholar
38. 1. Johnson DM
    2. Illig KR
    3. Behan M
    4. Haberly LB

    (2000)

    New features of connectivity in piriform cortex visualized by intracellular injection of pyramidal cells suggest that "primary" olfactory cortex functions like "association" cortex in other sensory systems

    The Journal of Neuroscience 20:6974–6982.

    • Google Scholar
39. 1. Karpuk N
    2. Hayar A

    (2008) Activation of postsynaptic GABAB receptors modulates the bursting pattern and synaptic activity of olfactory bulb juxtaglomerular neurons

    Journal of Neurophysiology 99:308–319.

    https://doi.org/10.1152/jn.01086.2007

    • Google Scholar
40. 1. Kreiman G
    2. Koch C
    3. Fried I

    (2000) Category-specific visual responses of single neurons in the human medial temporal lobe

    Nature Neuroscience 3:946–953.

    https://doi.org/10.1038/78868

    • Google Scholar
41. 1. Kriegeskorte N
    2. Mur M
    3. Bandettini P

    (2008) Representational similarity analysis - connecting the branches of systems neuroscience

    Frontiers in Systems Neuroscience 2:4.

    https://doi.org/10.3389/neuro.06.004.2008

    • Google Scholar
42. 1. Kriegeskorte N
    2. Mur M
    3. Ruff DA
    4. Kiani R
    5. Bodurka J
    6. Esteky H
    7. Tanaka K
    8. Bandettini PA

    (2008) Matching categorical object representations in inferior temporal cortex of man and monkey

    Neuron 60:1126–1141.

    https://doi.org/10.1016/j.neuron.2008.10.043

    • Google Scholar
43. 1. Kumaran D
    2. McClelland JL

    (2012) Generalization through the recurrent interaction of episodic memories: A model of the hippocampal system

    Psychological Review 119:573–616.

    https://doi.org/10.1037/a0028681

    • Google Scholar
44. 1. Lanthorn TH
    2. Cotman CW

    (1981) Baclofen selectively inhibits excitatory synaptic transmission in the hippocampus

    Brain Research 225:171–178.

    https://doi.org/10.1016/0006-8993(81)90326-7

    • Google Scholar
45. 1. LaRocque KF
    2. Smith ME
    3. Carr VA
    4. Witthoft N
    5. Grill-Spector K
    6. Wagner AD

    (2013) Global similarity and pattern separation in the human medial temporal lobe predict subsequent memory

    The Journal of Neuroscience 33:5466–5474.

    https://doi.org/10.1523/JNEUROSCI.4293-12.2013

    • Google Scholar
46. 1. Leutgeb S
    2. Leutgeb JK

    (2007) Pattern separation, pattern completion, and new neuronal codes within a continuous CA3 map

    Learning & Memory 14:745–757.

    https://doi.org/10.1101/lm.703907

    • Google Scholar
47. 1. Li W
    2. Howard JD
    3. Parrish TB
    4. Gottfried JA

    (2008) Aversive learning enhances perceptual and cortical discrimination of indiscriminable odor cues

    Science 319:1842–1845.

    https://doi.org/10.1126/science.1152837

    • Google Scholar
48. 1. Li W
    2. Luxenberg E
    3. Parrish T
    4. Gottfried JA

    (2006) Learning to smell the roses: Experience-dependent neural plasticity in human piriform and orbitofrontal cortices

    Neuron 52:1097–1108.

    https://doi.org/10.1016/j.neuron.2006.10.026

    • Google Scholar
49. 1. Logothetis NK
    2. Wandell BA

    (2004) Interpreting the BOLD signal

    Annual Review of Physiology 66:735–769.

    https://doi.org/10.1146/annurev.physiol.66.082602.092845

    • Google Scholar
50. Book

    1. Mai JK
    2. Assheuer J
    3. Paxinos G

    (1997)

    Atlas of the Human Brain

    Academic Press.

    • Google Scholar
51. 1. Nickell WT
    2. Behbehani MM
    3. Shipley MT

    (1994) Evidence for gabab-mediated inhibition of transmission from the olfactory nerve to mitral cells in the rat olfactory bulb

    Brain Research Bulletin 35:119–123.

    https://doi.org/10.1016/0361-9230(94)90091-4

    • Google Scholar
52. 1. Nili H
    2. Wingfield C
    3. Walther A
    4. Su L
    5. Marslen-Wilson W
    6. Kriegeskorte N

    (2014) A toolbox for representational similarity analysis

    PLoS Computational Biology 10:e1003553.

    https://doi.org/10.1371/journal.pcbi.1003553

    • Google Scholar
53. 1. Okutani F
    2. Zhang JJ
    3. Otsuka T
    4. Yagi F
    5. Kaba H

    (2003) Modulation of olfactory learning in young rats through intrabulbar GABA(B) receptors

    The European Journal of Neuroscience 18:2031–2036.

    https://doi.org/10.1046/j.1460-9568.2003.02894.x

    • Google Scholar
54. 1. Olofsson JK
    2. Hurley RS
    3. Bowman NE
    4. Bao X
    5. Mesulam MM
    6. Gottfried JA

    (2014) A designated odor-language integration system in the human brain

    The Journal of Neuroscience 34:14864–14873.

    https://doi.org/10.1523/JNEUROSCI.2247-14.2014

    • Google Scholar
55. 1. Palouzier-Paulignan B
    2. Duchamp-Viret P
    3. Hardy AB
    4. Duchamp A

    (2002) GABA(B) receptor-mediated inhibition of mitral/tufted cell activity in the rat olfactory bulb: A whole-cell patch-clamp study in vitro

    Neuroscience 111:241–250.

    https://doi.org/10.1016/s0306-4522(02)00003-9

    • Google Scholar
56. 1. Poo C
    2. Isaacson JS

    (2011) A major role for intracortical circuits in the strength and tuning of odor-evoked excitation in olfactory cortex

    Neuron 72:41–48.

    https://doi.org/10.1016/j.neuron.2011.08.015

    • Google Scholar
57. 1. Poppenk J
    2. Evensmoen HR
    3. Moscovitch M
    4. Nadel L

    (2013) Long-axis specialization of the human hippocampus

    Trends in Cognitive Sciences 17:230–240.

    https://doi.org/10.1016/j.tics.2013.03.005

    • Google Scholar
58. 1. Preston AR
    2. Shrager Y
    3. Dudukovic NM
    4. Gabrieli JD

    (2004) Hippocampal contribution to the novel use of relational information in declarative memory

    Hippocampus 14:148–152.

    https://doi.org/10.1002/hipo.20009

    • Google Scholar
59. 1. Riesenhuber M
    2. Poggio T

    (2000) Models of object recognition

    Nature Neuroscience 3 Suppl:1199–1204.

    https://doi.org/10.1038/81479

    • Google Scholar
60. Book

    1. Roach EH

    (1978)

    Cognition and Categorization

    Roach E. H, Lloyd B. B, editors. Erlbaum Associates.

    • Google Scholar
61. Website

    1. RxList The internet Drug Index

    (2007) Kemstro, side effects and drug interactions

    http://www.rxlist.com/kemstro-drug/side-effects-interactions.htm
62. 1. Schoenbaum G
    2. Eichenbaum H

    (1995)

    Information coding in the rodent prefrontal cortex. II. ensemble activity in orbitofrontal cortex

    Journal of Neurophysiology 74:751–762.

    • Google Scholar
63. 1. Seger CA
    2. Miller EK

    (2010) Category learning in the brain

    Annual Review of Neuroscience 33:203–219.

    https://doi.org/10.1146/annurev.neuro.051508.135546

    • Google Scholar
64. 1. Seger CA
    2. Peterson EJ

    (2013) Categorization = decision making + generalization

    Neuroscience and Biobehavioral Reviews 37:1187–1200.

    https://doi.org/10.1016/j.neubiorev.2013.03.015

    • Google Scholar
65. 1. Shakhawat AM
    2. Harley CW
    3. Yuan Q

    (2014) Arc visualization of odor objects reveals experience-dependent ensemble sharpening, separation, and merging in anterior piriform cortex in adult rat

    The Journal of Neuroscience 34:10206–10210.

    https://doi.org/10.1523/JNEUROSCI.1942-14.2014

    • Google Scholar
66. 1. Shohamy D
    2. Wagner AD

    (2008) Integrating memories in the human brain: Hippocampal-midbrain encoding of overlapping events

    Neuron 60:378–389.

    https://doi.org/10.1016/j.neuron.2008.09.023

    • Google Scholar
67. 1. Stettler DD
    2. Axel R

    (2009) Representations of odor in the piriform cortex

    Neuron 63:854–864.

    https://doi.org/10.1016/j.neuron.2009.09.005

    • Google Scholar
68. 1. Tang AC
    2. Hasselmo ME

    (1994) Selective suppression of intrinsic but not afferent fiber synaptic transmission by baclofen in the piriform (olfactory) cortex

    Brain Research 659:75–81.

    https://doi.org/10.1016/0006-8993(94)90865-6

    • Google Scholar
69. 1. Terrier J
    2. Ort A
    3. Yvon C
    4. Saj A
    5. Vuilleumier P
    6. Lüscher C

    (2011) Bi-directional effect of increasing doses of baclofen on reinforcement learning

    Frontiers in Behavioral Neuroscience 5:.

    https://doi.org/10.3389/fnbeh.2011.00040

    • Google Scholar
70. 1. Wachowiak M
    2. McGann JP
    3. Heyward PM
    4. Shao Z
    5. Puche AC
    6. Shipley MT

    (2005) Inhibition [corrected] of olfactory receptor neuron input to olfactory bulb glomeruli mediated by suppression of presynaptic calcium influx

    Journal of Neurophysiology 94:2700–2712.

    https://doi.org/10.1152/jn.00286.2005

    • Google Scholar
71. 1. Wilson DA

    (2009) Pattern separation and completion in olfaction

    Annals of the New York Academy of Sciences 1170:306–312.

    https://doi.org/10.1111/j.1749-6632.2009.04017.x

    • Google Scholar
72. 1. Wilson DA
    2. Sullivan RM

    (2011) Cortical processing of odor objects

    Neuron 72:506–519.

    https://doi.org/10.1016/j.neuron.2011.10.027

    • Google Scholar
73. 1. Wu KN
    2. Tan BK
    3. Howard JD
    4. Conley DB
    5. Gottfried JA

    (2012) Olfactory input is critical for sustaining odor quality codes in human orbitofrontal cortex

    Nature Neuroscience 15:1313–1319.

    https://doi.org/10.1038/nn.3186

    • Google Scholar
74. 1. Yassa MA
    2. Stark CE

    (2011) Pattern separation in the hippocampus

    Trends in Neurosciences 34:515–525.

    https://doi.org/10.1016/j.tins.2011.06.006

    • Google Scholar
75. 1. Young KA
    2. Franklin TR
    3. Roberts DC
    4. Jagannathan K
    5. Suh JJ
    6. Wetherill RR
    7. Wang Z
    8. Kampman KM
    9. O'Brien CP
    10. Childress AR

    (2014) Nipping cue reactivity in the bud: Baclofen prevents limbic activation elicited by subliminal drug cues

    The Journal of Neuroscience 34:5038–5043.

    https://doi.org/10.1523/JNEUROSCI.4977-13.2014

    • Google Scholar
76. 1. Ziakopoulos Z
    2. Brown MW
    3. Bashir ZI

    (2000) GABAB receptors mediate frequency-dependent depression of excitatory potentials in rat perirhinal cortex in vitro

    The European Journal of Neuroscience 12:803–809.

    https://doi.org/10.1046/j.1460-9568.2000.00965.x

    • Google Scholar

Author details

1. Xiaojun Bao

   Department of Neurology, Northwestern University Feinberg School of Medicine, Chicago, United States

   Contribution

   XB, Conception and design, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article

   For correspondence

   xiaojunbao2011@u.northwestern.edu

   Competing interests

   The authors declare that no competing interests exist.

   "This ORCID iD identifies the author of this article:" 0000-0002-8310-4141

2. Louise LG Raguet

   Department of Biology, École Normale Supérieure, Lyon, France

   Contribution

   LLGR, Conception and design, Acquisition of data

   Competing interests

   The authors declare that no competing interests exist.

3. Sydni M Cole

   Department of Neurology, Northwestern University Feinberg School of Medicine, Chicago, United States

   Contribution

   SMC, Conception and design, Acquisition of data

   Competing interests

   The authors declare that no competing interests exist.

4. James D Howard

   Department of Neurology, Northwestern University Feinberg School of Medicine, Chicago, United States

   Contribution

   JDH, Analysis and interpretation of data, Drafting or revising the article

   Competing interests

   The authors declare that no competing interests exist.

5. Jay A Gottfried

   1. Department of Neurology, Northwestern University Feinberg School of Medicine, Chicago, United States
   2. Department of Psychology, Northwestern University Weinberg College of Arts and Sciences, Evanston, United States

   Contribution

   JAG, Conception and design, Analysis and interpretation of data, Drafting or revising the article

   For correspondence

   j-gottfried@northwestern.edu

   Competing interests

   The authors declare that no competing interests exist.

• 2,287

  views

• 425

  downloads

• 20

  citations

Views, downloads and citations are aggregated across all versions of this paper published by eLife.