, 2010), consistent with their role as signaling a tonic modulatory tone. Moreover, tonic low concentrations of DA seem to be important for maintaining circuit basal function, while

phasic, higher concentrations produce shorter-term modulation (Rodgers et al., 2011a, 2011b). One of the most puzzling questions arising from extensive neuromodulation is how the integrity of the modulated circuits is maintained, although so may circuit parameters can be altered. If one tries to build a computational model of either a single neuron or a circuit, it can be quite hard to find a set of parameters that are consistent with the CHIR-99021 concentration desired output. Indeed, random assignment of parameters to a single neuron or a circuit will lead to significantly more failures than successful models (Prinz, 2010; Prinz et al., 2003a, 2004; Taylor et al., 2009). Nonetheless, there are many different sets of parameters that can produce similar output patterns (Goldman et al., 2001; Prinz et al., 2004; Taylor et al., 2009). There are circumstances in which neuromodulators are used to qualitatively transform the behavior of a circuit, such as during transitions from sleep to wakefulness (McCormick, 1989, 1992; McCormick and Bal, 1997) or when a hormonal pathway is used to trigger eclosion (Kim et al., 2006) or molting (Webster et al., 2012). There are also neuromodulatory influences

that reshape networks during ongoing behavior, and the sets of parameters KU-55933 molecular weight that are produced by neuromodulator action must be consistent with stable and appropriate cellular Florfenicol and circuit function (Goldman et al., 2001). Understanding how circuits can be stable in the face of ubiquitous neuromodulation is an important and deep problem. Why don’t the circuits important for behavior become “overmodulated” more often, and what mechanisms might protect against overmodulation? The answers to this question may be partially idiosyncratic

to each circuit, but I suggest some general mechanisms that may play a role in maintaining functional circuit performance during modulation. Harris-Warrick and Johnson (2010) suggest that the pattern of dopamine modulation of STG neurons at the cellular level (Figure 5) is ideally suited to maintain stable function. Specifically, by acting on both inward and outward currents, dopamine actions can keep individual neurons, and therefore the network, within their operating range (Harris-Warrick and Johnson, 2010). The importance of the voltage dependence of the NMDA receptor for the induction of LTP is well appreciated, but the ability of the NMDA receptor to induce oscillations in the spinal cord is less well known (Sigvardt et al., 1985). The neuropeptide proctolin elicits a voltage-dependent inward current similar to that evoked by NMDA (Golowasch and Marder, 1992). This current is blocked at hyperpolarized membrane potentials by extracellular Ca2+ and has a reversal potential about 0mV.

diff), and clustering coefficient of genes annotated by terms with the highest TS scores were compared to the rest of the module, allowing us to home in on particularly tight-knit, Selleck Cabozantinib behaviorally relevant, biological pathways/functions in the singing-related modules (Supplemental Experimental Procedures). For example, 11 genes in the blue module (ARC, CABP1, CNN3, DLG1, DLG2, DLGAP2, FREQ, HOMER1, IFNGR1, NLGN1, and NTRK2) were annotated by the term “GO:0014069∼postsynaptic density” ( Table S4). Probes representing these genes in the blue module had high MM and GS.motifs.X (27 probes

total; mean MM = 0.804, GS.motifs.X = 0.682), and the term “GO:0014069∼postsynaptic density” had an enrichment p value of 0.059. Thus TS for this term = 0.804 × 0.682 × (1 − 0.059) = 0.516 (7th highest selleck of 402 enriched blue module terms; Table S2. Area X Network Data and Table S4. Functional Annotation of Selected Modules). Compared to the rest of the module, probes for the 11 genes annotated with this term had higher average MM (p = 6.2e-7, Kruskal-Wallis test), GS.motifs.X (p = 6.8e-5), kIN.diff (p = 4.7e-6), and clustering coefficient (p = 5.2e-5). Other top-ranked blue module terms included “GO:0031434∼mitogen-activated protein kinase kinase binding” and “IPR019583:PDZ-associated domain of

NMDA receptors,” as well as others involving actin, cytoskeleton, and tyrosine phosphatase regulation. Genes associated with these synapse-related functions in the blue module were also some of FOXP2′s closest neighbors, i.e., genes with which it had high TO ( Figures 6D–6F, Table S2, Supplemental Experimental Procedures). This may imply a role for FoxP2 in the Cytidine deaminase suppression of synaptic plasticity, since blue module genes (whose levels increased with singing in these experiments) in high TO with FOXP2 (which decreased with singing) are good candidates for repressed transcriptional targets. Each of the song modules was enriched for astrocytic markers with developing astrocytes most enriched in the blue module (p = 7.5e-6, Fisher’s

exact test) and mature astrocytes in the orange module (p = 4e-3; Cahoy et al., 2008). This observation is consistent with the recent realization that astrocytes are involved in the regulation of neuronal functions, including behavior (Halassa and Haydon, 2010). We screened the modules for genes associated with Parkinson’s disease (Supplemental Experimental Procedures), since it is a basal ganglia based disorder with a vocal component and found enrichment in the black singing-related module (Figure S6). Another module that was moderately singing-related was also enriched for Parkinson’s disease-associated genes, as well as autism susceptibility genes (purple module, p = 2.7e-4, p = 0.05, respectively, Table S2). The unique presence of the song modules in area X implies that the biological pathways they represent are coregulated in patterns specific to area X during learned vocal-motor behavior.

Type II TARPs are more distantly related to type I and share only some of their functional properties ( Kato et al., 2010). Recent genetic and proteomic screens have identified a number of small proteins that bind to AMPARs and are structurally unrelated to TARPs (Figure 3). These include

cornichon-2 and -3 (CNIH-2 and CNIH-3) (Schwenk et al., 2012 and Schwenk et al., 2009), CKAMP44 (von Engelhardt et al., 2010), SynDIG1 (Kalashnikova et al., 2010), selleck chemicals GSG1L (Shanks et al., 2012), and in C. elegans SOL-1 and SOL-2 ( Wang et al., 2012). The most studied of these proteins are CNIH proteins, which profoundly slow the deactivation of AMPARs in heterologous systems ( Coombs et al., 2012, Gill et al., 2012, Schwenk et al., 2009 and Shi et al., 2010). Genetic deletion of CNIH-2 and -3 together causes a profound and selective see more loss of synaptic and surface AMPARs in the hippocampus ( Herring et al., 2013). This deficit is due to the selective loss of surface GluA1-containing AMPARs (GluA1/A2 heteromers), leaving a small residual pool of synaptic GluA2/A3 heteromers. The kinetics

of AMPARs in neurons lacking CNIH-2/-3 are faster than those in WT neurons due to the fast kinetics of GluA2/A3 heteromers. The remarkably selective effect of CNIHs on the GluA1 subunit appears to be mediated by TARP γ-8, which prevents a functional association of CNIHs with non-GluA1 subunits. Surprisingly, although CNIHs strongly slow deactivation in heterologous cells, they do not directly affect the kinetics of surface neuronal AMPARs, indicating either that they dissociate from the AMPARs in the Golgi/ER akin to the chaperoning role of their yeast and Drosophila homologs or that their selective binding to surface GluA1 subunits of GluA1/A2 heteromers is functionally silent. These results point to a sophisticated

interplay between CNIHs and TARP γ-8 that dictates subunit-specific AMPAR trafficking and the strength and kinetics of synaptic AMPAR-mediated transmission. CKAMP44 is expressed at high levels in dentate granule cells where it enhances AMPAR desensitization and recovery from desensitization, Tolmetin thus impacting short-term plasticity ( von Engelhardt et al., 2010). Neuroligins (NLs) and leucine-rich repeat transmembrane proteins (LRRTMs) are postsynaptic adhesion molecules that bind to presynaptic neurexins and are involved in excitatory synapses assembly, maturation, and specification (Craig and Kang, 2007, Krueger et al., 2012 and Südhof, 2008) (Figure 3). However, recent findings indicate that both NLs and LRRTMs have more specific roles in both AMPAR trafficking and LTP. Knockdown of LRRTM1 and LRRTM2 in CA1 neurons selectively reduces AMPAR-EPSCs in the neonate (Soler-Llavina et al., 2011), although in dentate granule cells the NMDAR-EPSC is also reduced (de Wit et al., 2009).

, 2001). Furthermore, the mutated calcineurin gene (PPP3CC) has been shown to map to chromosomal loci previously implicated in schizophrenia by genetic linkage studies ( Gerber et al., 2003). Taken together, these features suggest that the calcineurin KO provides a unique opportunity to investigate the neural basis of UMI-77 cell line dysfunction in a schizophrenia model. The hippocampus is a brain structure critical for episodic memory (Gaffan, 1994, Olton and Samuelson, 1976, Scoville and Milner, 1957 and Steele and Morris, 1999) and spatial learning (Morris et al., 1982 and O’Keefe and Nadel, 1978). In freely moving rodents, the hippocampus

exhibits distinct activity profiles dependent on behavioral state (Buzsáki, 1989), suggesting distinct modes of information processing within the structure. During running, the hippocampal electroencephalogram (EEG) exhibits a 4–12 Hz theta rhythm (Skaggs et al., 1996), and hippocampal

principal neurons exhibit location-specific responses, known as place fields, as reported in rats (O’Keefe and Dostrovsky, 1971), mice (McHugh et al., 1996), monkeys (Matsumura et al., 1999), and humans (Ekstrom et al., 2003). By contrast, during awake rest periods, hippocampal EEG is distinguished by sharp-wave-ripple (SWR) events (Buzsáki, 1989) and hippocampal principal neurons take part in extended sequences of coactivity, which replay previous behavioral episodes Enzalutamide datasheet (Davidson et al., 2009, Diba and Buzsáki, 2007, Foster and Wilson, 2006 and Gupta et al., 2010) as well as preplay subsequent behavioral episodes (Dragoi and Tonegawa, 2011, Dragoi and Tonegawa, 2013 and Pfeiffer and Foster, 2013). There is substantial evidence linking schizophrenia with damage to the hippocampus (Weinberger, 1999). Dysfunction of the hippocampus and related medial temporal lobe structures has also been reported in schizophrenia patients (Small

et al., 2011), together with selective impairments in learning and memory. In addition, abnormal brain activity in schizophrenia patients has been detected in various brain structures, including mafosfamide the hippocampus, during rest periods (Buckner et al., 2008) and during passive task epochs (Harrison et al., 2007). Since the pattern of impairments of calcineurin KO mice—synaptic plasticity changes in the hippocampus and hippocampal-dependent behavioral phenotypes such as working memory—suggested that hippocampal function might be affected in this mouse model of schizophrenia, we targeted the hippocampus for electrophysiological recordings in freely behaving KO and littermate controls (CT) and investigated changes in information processing during exploratory behavior and resting periods. To characterize hippocampal activity in our mouse model, we employed microdrives with multiple independently adjustable tetrodes to record single-unit spikes and EEG from the CA1 subregion of the dorsal hippocampus of freely behaving KO mice (n = 7) and floxed littermate CT (n = 5).

Finally, it is notable that of the two major ePN targets, iPN axons only project to the lateral horn but spare the mushroom body (Figure 7D). The mushroom body is a well-documented center for olfactory learning and memory, whereas PN projections selleck chemical to the lateral horn are implicated in regulating innate

olfactory behavior (Heimbeck et al., 2001) (see also Parnas et al., 2013). ePN axons exhibit striking stereotypy in their terminal arborization patterns in the lateral horn, but not in the mushroom body (Caron et al., 2013, Jefferis et al., 2007, Marin et al., 2002 and Wong et al., 2002). Recent anatomical tracing in mice also revealed differential input organization in distinct olfactory cortical areas (Miyamichi

et al., 2011 and Sosulski selleck kinase inhibitor et al., 2011), suggesting a common principle in olfactory systems of insects and mammals. The selective innervation by iPNs of targeting neurons in the lateral horn suggests that regulation of innate olfactory behavior engages an additional level of specific inhibition to ensure that olfactory information carrying different biological values, such as food and pheromone, is funneled into distinct downstream circuits, resulting in the activation of distinct behavioral outputs. Two-photon GCaMP imaging experiments were performed with either a LSM 510 Two-Photon Laser-Scanning Confocal Microscope (Zeiss) with a 40× NA 0.8 water-immersion objective (Zeiss) and modelocked Ti:Sapphire laser (Coherent) tuned to 920 nm or a customized two-photon microscope (Prairie Technologies) with a 40× NA 1.0 water-immersion all objective (Zeiss) and laser tuned to 927 nm at ∼73–75°F. The excitation power at the specimen was ∼10 mW, and the pixel dwell time was 2.0 μs. All lateral horn images were acquired at a 2.488 Hz frame rate with 460 × 300 pixels per frame. Each imaging cycle was 45 s, and

500 msec odor stimuli (as determined by the solenoid valves) were always delivered at 5 s. To minimize bleaching, images were only taken from the first 16 s (40 frames) of each cycle. In most experiments, the same odor was applied every other cycle for three repeats, while different odors were usually applied in an alternate manner to minimize potential olfactory adaptation. Image acquisition was suspended during the 500 msec optogenetic stimulation period to protect the PMTs. On average, each imaging session lasted ∼1.5 hr, with most of the flies appearing healthy at the end of the experiments; they could still move their legs at a regular pace. For some experiments, the fly brains were dissected and fixed for post hoc staining. Time-lapse imaging series of GCaMP3 from a single z plane were usually recorded in the control hemisphere once before mACT transection and once after transection.

Transient heterosynaptic suppression driven by strong PFC activity may facilitate transmission of PFC-related information by the VS through basal ganglia loops. Whereas HP inputs may subserve a critical gating function, the impact of burst-like PFC activity upon information processing in the VS is clearly distinct from that of HP activity. Behavioral studies indicate different functional impact of PFC and HP inputs to the VS. For example, whereas limbic afferents to the VS readily elicit self-stimulation behavior, similar PFC stimulation fails to do so (Stuber et al., 2011). More recently, optical stimulation of PFC afferents to the VS were found

to be reinforcing in mice (Britt et al., 2012); however, in this case self-stimulation behavior required greater frequency and duration stimuli for CP-673451 order PFC than HP or amygdala inputs to be effective. These findings suggest that cortical inputs may have a qualitatively different connectivity in VS circuits than HP inputs and that responses to convergent PFC and HP inputs may not be additive in the VS. We propose that suppression of HP responses by strong PFC activation may allow an efficient transfer of PFC commands through AZD6738 price basal ganglia loops and an unhindered selection of the appropriate behavioral response. As the role of thalamic inputs

to the VS is not well understood, the functional implications of the PFC-thalamic input interaction are unclear. Thalamic afferents arriving to striatal regions primarily originate in the nonspecific nuclei (Groenewegen and Berendse, 1994). These projections are therefore likely to be involved in a global-activating function and perhaps in conveying crude sensory information. Transient suppression of this influence by strong PFC activation may facilitate the relay of PFC information through the VS with minimal disturbance from ongoing arousal state-related information. The impact of bursts of PFC activity on VS physiology may be essential for supporting cognitive functions that depend on the PFC. The VS itself is critical for instrumental

Edoxaban behavior and is required for the normal ability of animals to choose delayed reward (Cardinal et al., 2002). Furthermore, a distributed subset of VS neurons becomes active during decision points in a spatial navigation task (van der Meer and Redish, 2009). PFC-VS interactions are critical for rodent decision making (Christakou et al., 2004; St Onge et al., 2012) but are also important for human cognition. Deep electroencephalogram recordings during a reward-based learning task in humans reveal brief epochs of synchronous activity in the VS and medial PFC during decision-making instances (Cohen et al., 2009). In addition to transiently enhanced PFC-VS activity, several studies indicate that interactions between the HP and VS vary during epochs that require decisions. Simultaneous local field potential recordings from both structures reveal that ventral HP-VS coupling is altered during performance of a T-maze task (Tort et al.

The brain was quickly removed and

placed in ice-cold slicing solution containing (in mM) 87 NaCl, 2.5 KCl, 25 NaHCO3, 0.5 CaCl2, 7 MgCl2, 1.25 NaH2PO4, 25 glucose, and 75 sucrose saturated with 95%/5% O2/CO2. The brain was blocked and mounted on a vibrating slicer (Leica, Nussloch, Germany) submerged in ice-cold slicing solution. Angled horizontal slices (250 μm) containing the DMH were cut and incubated in 32.5°C artificial cerebrospinal fluid (aCSF) containing (in mM) 126 NaCl, 2.5 KCl, 26 NaHCO3, 2.5 CaCl2, 1.5 MgCl2, AT13387 solubility dmso 1.25 NaH2PO4, and 10 glucose saturated with 95%/5% O2/CO2 for a minimum of 60 min. Hypothalamic slices were then submerged in a recording chamber and superfused with 32.5°C aCSF at a flow rate of 1 ml/min. Whole-cell recordings were obtained from DMH neurons visualized with an Olympus upright microscope (Olympus, Center Valley, PA) fitted with infrared differential interference contrast optics. Recordings were obtained using borosilicate glass microelectrodes (tip resistance 4.5–6.5 MΩ) filled with a solution containing (in mM) 108 K gluconate, 8 Na gluconate, 2 MgCl2, 8 KCl, 1 potassium EGTA, 4 potassium ATP, 0.3 sodium GTP, and 10 HEPES, and corrected to pH 7.2 with

KOH. In a subset of experiments, 10 mM BAPTA was included in the intracellular solution to chelate postsynaptic Ca2+. www.selleckchem.com/products/abt-199.html Recordings were accepted for analysis if changes in access resistance were <15%. Cells were voltage clamped at −80 mV and the perfusate always contained DNQX (10 μM; Tocris, Ellisville, MO) to block AMPA and kainate receptor-mediated glutamatergic transmission. GABAergic fibers were stimulated extracellularly with a patch pipette filled with aCSF and positioned within

the DMH. Oxalosuccinic acid IPSCs were evoked at a rate of 0.2 Hz and paired-pulse responses were obtained by applying a pair of synaptic stimuli 50 ms apart. For HFS, afferents were stimulated at 100 Hz for 4 s, repeated twice, 20 s apart, unless otherwise specified. Electrophysiological signals were amplified using the Multiclamp700 B amplifier (Molecular Devices, Union City, CA), low-pass-filtered at 1 kHz, digitized at 10 kHz using the Digidata 1322 (Molecular Devices), and stored for offline analysis. Evoked currents were analyzed using Clampfit 9 (Molecular Devices). The amplitude of the synaptic current was calculated from the baseline (current before evoked response) to the peak of each evoked. For clarity, the stimulus artifacts were removed digitally from the traces depicted. Spontaneous IPSCs were analyzed using the threshold detection criteria in Minianalysis (Synaptosoft). Results are expressed as means ± SEM. In most cases, significance was determined using a one-sample or paired Student’s t test comparing the means following HFS or drug treatment to baseline with significance level of p < 0.05.

36 However, as reported by Arnason et al.,38 it was not possible to identify football-specific buy Everolimus screening tests to identify an increased risk of ankle sprain pre-injury, apart from having sustained a previous ankle strain. This study revealed on AT a tendency towards an increased dorsiflexion angle at touchdown, a trend towards higher external rotation at weight acceptance and for the 30° cut an increased inversion at the beginning

and end of the early acceleration phase. Hence, no clear strategy to support or refute increased ankle injury risk derived out of this study, and further research is needed to fully understand the surface–player effect on the ankle joint. The current study has shown surface-induced alterations occurred in the kinematics of female football players, a more in depth analysis including ground reaction forces, joint kinetics, and EMG could reveal additional information and increase our understanding of the interaction between the female

player and the different surface systems in football-specific situations. It has to be noted that a variety of 3G AT systems exists and the differences in movement between ATs could become greater than between AT and NT.2 Therefore the results of this study can only be applied to the differences ZD1839 between the specific AT and NT used. Athletes wore the same football boot, which they would wear on both surfaces, which might not be the football boot used in match play. However, boot type (studded vs. bladed) did not seem to impart differences in knee loading when used on AT, 39 and this approach allowed an investigation on surface-induced rather than shoe-induced effects. As the movement changes induced by AT are not well understood, and gender related responses might be affected by a variety

of different aspects, such as climatic exposure, boot Chlormezanone type, or playing experience, a number of key research questions remain unanswered, and our understanding of the influence of artificial surfaces needs to be further developed. These investigations should address more factorial approaches as including males and different soccer relevant movements (e.g., straight running vs. cutting with different angles). Finally, the present study investigated only a small sample size, as such, the findings should be interpreted with care and only can point out tendencies towards the discussed kinematic changes. Using a higher sample size could possibly lead to not only similar or decreased effect sizes, but also current non-significant differences could become significant. The overall purpose of this study was to investigate the lower limb kinematics on different surfaces in female football players during an unanticipated cutting manoeuvre. The major finding of this study was that there was no evidence to suggest that there is an increased risk of injury when performing with the same movement speed on an AT.

MGE cells migrating on cortical axons were sectioned parallel to the plane of

migration (Figures 1D1 and 1D2). Semithin sections comprising both the CTR and the nucleus were analyzed using high-resolution electron tomography (Koster Vismodegib datasheet et al., 1997). In a large proportion of cells with long nucleus to CTR distances the mother centriole identified by the presence of lateral and/or distal appendages was associated to the plasma membrane by its distal end (Figures 1E–1F2 and 1L; 21 cells out of 33). A third of these cells had a short primary cilium that protruded from the mother centriole into the extracellular space. This primary cilium contained an axoneme (Figures 1F1 and 1F2 and Movie S1) and was often less than 500 nm in length, shorter than the primary cilium found on fully differentiated neurons of adult brains (Fuchs and Schwark, 2004; Arellano et al., 2012). The plasma membrane around the primary cilium often formed a thickened asymmetric depression (Figure 1F1). Mother centrioles located in the leading process often associated with the plasma membrane. In contrast, centriole pairs located in the perinuclear compartment positioned deep within the cytoplasm (Figures 1G–1I, 1L, S1C, and S1D). There, the mother centriole associated with a large distal vesicle, either round or flattened (Figures 1H and 1I and Movie S2). A short axoneme could protrude

from the mother centriole within the vesicle lumen (Figure 1I, black arrow heads). The single large vesicle

was sometimes replaced Autophagy Compound Library by a row of small vesicles attached to the tip of mother centriole distal appendages (Figure S1D). Pioneer studies (Sorokin, Olopatadine 1962; Cohen et al., 1988) already reported that the ciliogenesis likely starts with the assembly of a centriolar vesicle into which the axoneme elongates. The centriolar vesicle of MGE cells could engulf smaller vesicles (Figure 1H and Movie S2), attesting to vesicular trafficking toward the centriolar vesicle. Accordingly, we noticed a continuum of small vesicles between the neighboring Golgi cisternae and the large centriolar vesicle (Figure 1I, white arrow heads). To obtain further insight into ciliogenesis related vesicular trafficking in migrating MGE cells, we examined the distribution of GMAP-210, a cis-Golgi protein that traffics toward the basal body in ciliated cells ( Ríos et al., 2004) and that associates with IFT20 ( Follit et al., 2008), a component of anterograde IFT particles. The cis-GA, as decorated by GMAP-210 antibodies, extended to the CTR, which was not the case for the median GA ( Figures 1J and S1E1–S1E3). A GMAP-210 positive Golgi compartment remained associated to the CTR after brefeldin treatment that redistributed the Golgi to the ER but not after MT destabilization ( Figures S1F1–S1G2).

Thus, the L1 norm is called sparse, and the corresponding neural representation is sparse overcomplete. It was shown that the recurrent network of inhibitory neurons can implement sparse overcomplete representations (Rozell et al., 2008). To show this, the network dynamics are represented as a minimization of a cost function called the Lyapunov function, similarly to the representation of Hopfield networks (Hertz et al., 1991 and Hopfield, 1982). Hopfield

networks have attractor states that contain memory of activation patterns. In contrast to Hopfield networks, in purely http://www.selleckchem.com/products/ulixertinib-bvd-523-vrt752271.html inhibitory networks, the recurrent weights enter the Lyapunov function with a minus sign, which abolishes the attractor memory states and makes the NVP-BKM120 mouse network purely sensory (Rozell et al., 2008). Minimization of Lyapunov function in realistic recurrent

networks with inhibition was suggested as a means to implement the parsimony constraint (L1) mentioned above. To implement sparse overcomplete representations with realistic networks of neurons, two requirements have to be met (Rozell et al., 2008). First, the feedforward weights between the input layer of the network and the inhibitory neurons have to contain the dictionary elements (Figure 8A). This ensures that inhibitory neurons representing a particular dictionary element will be driven strongly when it is present in the input, due to a high overlap between the stimulus and the feedforward weights. Second, the recurrent inhibitory weight between any pair of neurons has to be proportional to the overlap between their dictionary elements (Figure 8A). This feature implies that similarly through tuned inhibitory neurons compete more strongly. Therefore, the two types of network weights, feedforward and recurrent, have to closely match each other, one of them constructed as the overlap of the other. Here, we suggest that the olfactory bulb network architecture based on dendrodendritic synapses can ensure that the feedforward and recurrent connectivity are closely matched. In the architecture

based on dendrodendritic synapses, both the feedforward weights received by the GCs and their recurrent connections are dependent on the same set of synapses. Similar architectures have been proposed for analysis-synthesis networks (Mumford, 1994 and Olshausen and Field, 1997). The GCs of the olfactory bulb receive excitatory inputs from the MCs through dendrodendritic synapses (Shepherd et al., 2004). These synapses encode patterns that can strongly drive individual GCs. The effective connectivity between GCs is inhibitory (GC-to-MC and MC-to-GC synapses are inhibitory and excitatory, respectively). To calculate the strength of mutual inhibition, one has to calculate the sum over intermediate synapses, which leads to the evaluation of a convolution or overlap between GC input weights (Figure 8B).