Uma R. Karmarkar

Timing in the Absence of Clocks: Encoding Time in Neural Network States

Decisions based on the timing of sensory events are fundamental to sensory processing. However, the mechanisms by which the brain measures time over ranges of milliseconds to seconds remain unclear. The dominant model of temporal processing proposes that an oscillator emits events that are integrated to provide a linear metric of time. We examine an alternate model in which cortical networks are inherently able to tell time as a result of time-dependent changes in network state. Using computer simulations we show that within this framework, there is no linear metric of time, and that a given interval is encoded in the context of preceding events. Human psychophysical studies were used to examine the predictions of the model. Our results provide theoretical and experimental evidence that, for short intervals, there is no linear metric of time, and that time may be encoded in the high-dimensional state of local neural networks.

Experience-Dependent Plasticity in Adult Visual Cortex

Experience-dependent plasticity is a prominent feature of the mammalian visual cortex. Although such neural changes are most evident during development, adult cortical circuits can be modified by a variety of manipulations, such as perceptual learning and visual deprivation. Elucidating the underlying mechanisms at the cellular and synaptic levels is an essential step in understanding neural plasticity in the mature animal. Although developmental and adult plasticity share many common features, notable differences may be attributed to developmental cortical changes at multiple levels. These range from shifts in the molecular profiles of cortical neurons to changes in the spatiotemporal dynamics of network activity. In this review, we will discuss recent progress and remaining challenges in understanding adult visual plasticity, focusing on the primary visual cortex.

Book Review: How Do We Tell Time?

Animals time events on scales that range more than 10 orders of magnitude—from microseconds to days. This review focuses on timing that occurs in the range of tens to hundreds of milliseconds. It is within this range that virtually all the temporal cues for speech discrimination, and haptic and visual processing, occur. Additionally, on the motor side, it is on this scale that timing of fine motor movements takes place. To date, psychophysical data indicate that for many tasks there is a centralized timing mechanism, but that there are separate networks for different intervals. These data are supported by experiments that show that training to discriminate between two intervals generalizes to different modalities, but not different intervals. The mechanistic underpinnings of timing are not known. However various models have been proposed, they can be divided into labeled-line models and population clocks. In labeled-line models, different intervals are coded by activity in independent and discrete populations of neurons. In population models, time is coded by the population activity of a large group of neurons, and timing requires dynamic interaction between neurons. Population models are generally better suited for parallel processing of interval, duration, order, and sequence cues and are thus more likely to underlie timing in the range of tens to hundreds of milliseconds.

Different forms of homeostatic plasticity are engaged with distinct temporal profiles.

Global changes in network activity have been reported to induce homeostatic plasticity at multiple synaptic and cellular loci. Though individual types of plasticity are normally examined in isolation, it is their interactions and net effect that will ultimately determine their functional consequences. Here we examine homeostatic plasticity of both inhibition and intrinsic excitability in parallel in rat organotypic hippocampal slices. As previous studies have not examined inhibitory plasticity using a functional measure, inhibition was measured by the ability of evoked inhibitory postsynaptic potentials (IPSPs) to suppress action potentials, as well as IPSP amplitude. We show that manipulations of network activity can both up‐ and downregulate functional inhibition, as well as intrinsic excitability. However, these forms of plasticity are dissociable. Specifically, robust changes in intrinsic excitability were observed in the absence of inhibitory plasticity, and shifts in inhibition, but not excitability, appear to be sensitive to developmental stage. Our data establish that while the two forms of homeostatic plasticity can be engaged in parallel, there is a specific order in which they are expressed, with changes in excitability preceding those in inhibition. We propose that changes in intrinsic excitability occur first in order to stabilize network activity while optimizing the preservation of information stored in synaptic strengths by restricting changes that will disrupt the balance of synaptic excitation and inhibition.

Mechanisms and significance of spike-timing dependent plasticity

Abstract. Hebbs original postulate left two important issues unaddressed: (i) what is the effective time window between pre- and postsynaptic activity that will result in potentiation? and (ii) what is the learning rule that underlies decreases in synaptic strength? While research over the past 2 decades has addressed these questions, several studies within the past 5 years have shown that synapses undergo long-term depression (LTD) or long-term potentiation (LTP) depending on the order of activity in the pre- and postsynaptic cells. This process has been referred to as spike-timing dependent plasticity (STDP). Here we discuss the experimental data on STDP, and develop models of the mechanisms that may underlie it. Specifically, we examine whether the standard model of LTP and LTD in which high and low levels of Ca2+ produce LTP and LTD, respectively, can also account for STDP. We conclude that the standard model can account for a type of STDP in which, counterintuitively, LTD will be observed at some intervals in which the presynaptic cell fires before the postsynaptic cell. This form of STDP will also be sensitive to parameters such as the presence of an afterdepolarization following an action potential. Indeed, the sensitivity of this type of STDP to experimental parameters suggests that it may not play an important physiological role in vivo. We suggest that more robust forms of STDP, which do not exhibit LTD at pre–before–post intervals, are not accounted for by the standard model, and are likely to rely on a second coincidence detector in addition to the NMDA receptor.

Evaluating dedicated and intrinsic models of temporal encoding by varying context

Two general classes of models have been proposed to account for how people process temporal information in the milliseconds range. Dedicated models entail a mechanism in which time is explicitly encoded; examples include clock–counter models and functional delay lines. Intrinsic models, such as state-dependent networks (SDN), represent time as an emergent property of the dynamics of neural processing. An important property of SDN is that the encoding of duration is context dependent since the representation of an interval will vary as a function of the initial state of the network. Consistent with this assumption, duration discrimination thresholds for auditory intervals spanning 100 ms are elevated when an irrelevant tone is presented at varying times prior to the onset of the test interval. We revisit this effect in two experiments, considering attentional issues that may also produce such context effects. The disruptive effect of a variable context was eliminated or attenuated when the intervals between the irrelevant tone and test interval were made dissimilar or the duration of the test interval was increased to 300 ms. These results indicate how attentional processes can influence the perception of brief intervals, as well as point to important constraints for SDN models.

Cost Conscious? The Neural and Behavioral Impact of Price Primacy on Decision-Making

Price is a key factor in most purchases, but it can be presented at different stages of decision making. The authors examine the sequence-dependent effects of price and product information on the decision-making process at both neural and behavioral levels. During functional magnetic resonance imaging, the price of a product was shown to participants either before or after the product itself was presented. Early exposure to price, or “price primacy,” altered the process of valuation, as observed in altered patterns of activity in the medial prefrontal cortex immediately before making a purchase decision. Specifically, whereas viewing products first resulted in evaluations strongly related to products’ attractiveness or desirability, viewing prices first appeared to promote overall evaluations related to products’ monetary worth. Consistent with this framework, the authors show that price primacy can increase purchase of bargain-priced products when their worth is easily recognized. Together, these results suggest that price primacy highlights considerations of product worth and can thereby influence purchasing.

Asymmetric Effects of Favorable and Unfavorable Information on Decision Making Under Ambiguity

Most daily decisions involve uncertainty about outcome probabilities arising from incomplete knowledge, i.e., ambiguity. We explore how the addition of partial information affects these types of choices using theoretical and empirical methods. Our experiments in both gain and loss domains demonstrate that when such information supports a favorable outcome, it strongly increases valuation of an ambiguous financial prospect. However, when information supports an unfavorable outcome, it has significantly less impact. We find that two mechanisms drive this asymmetry. First, unfavorable information decreases estimates of a good outcome occurring but also reduces aversive uncertainty. These factors act in opposition, minimizing the effects of unfavorable information. Second, when information can be subjectively interpreted, unfavorable information is less likely to be integrated into evaluations. Our findings reveal mechanisms not captured by traditional models of decision making under uncertainty and highlight the importance of increasing the salience of unfavorable information in uncertain contexts to promote unbiased decision making.Data, as supplemental material, are available at http://dx.doi.org/10.1287/mnsc.2015.2233 . This paper was accepted by Teck-Hua Ho, behavioral economics .

Customer Experience and Service Design

While services already dominate economic activity in all major economies in the world, there has been curiously little investigation into many aspects of service management. For example, while product design and development have received a great deal of attention, the subject of service design has not been very visible in the research literature. There are many individual designers and design firms famous for their contributions to product design, but the same cannot be said for services. Undoubtedly many examples of outstanding service design exist and we will mention some later in this work. But recognition of service design as a discipline, as a management function or a job description, still seems to be rare.

Defining the Contributions of Network Clock Models to Millisecond Timing

Our ability to measure time extends from microseconds to days (Buonomano and Karmarkar, 2002; Buhusi and Meck, 2005). Given the wealth of experimental support for a number of different models of timing, it has been recently suggested that multiple mechanisms act in concert to transition smoothly between both temporal ranges and modalities (Wiener et al., 2011). This underscores the importance of determining the specific contributions of individual mechanisms to better understand these transitions and any relevant idiosyncrasies of timing in a particular context. We focus here on interval discrimination in the range of tens to hundreds of milliseconds, which plays an important role in a variety of tasks, such as speech processing, motion detection, and fine motor coordination. It has been proposed that such timing can emerge directly from the temporal properties intrinsic to neural circuits (Buonomano and Mauk, 1994; Buonomano and Merzenich, 1995; Karmarkar and Buonomano, 2007; for review see Ivry and Schlerf, 2008). Broadly, this class of mechanisms can be described as population or network clocks, as the timing of incoming stimuli is coded as the changes they effect in a population of neurons. This can also be thought of as the change in the overall state of the network. Compared to models of millisecond timing based on a single centralized mechanism (Treisman, 1963; Church, 1984; Gibbon et al., 1997; Ivry and Spencer, 2004) network clock models imply that timing is being done in multiple loci across the cortex. This is interesting because it means that the dynamics of timing are dependent on the properties of the underlying local circuitry, which could vary across modalities. To understand the behavior of network clocks, it is useful to examine a particular instantiation, referred to as a state-dependent network (SDN), which has been developed in the context of sensory processing (Karmarkar and Buonomano, 2007). For SDNs an incoming stimulus changes the network state not only by causing some population of neurons to fire, but also by engaging a number of intrinsic properties such as short-term plasticity, that change with specific time constants on the millisecond scale. As a result, the response of the network to a particular piece of temporal information is dependent on its recent history. Thus instead of marking each interval separately, timing is done continuously, with the network linking multiple signals together as a temporal object. The SDN can only measure information independently, or reset, when the interval between stimuli has been sufficiently long to allow the network to return to its baseline state. This makes two predictions about interval discrimination, the first being that a variable context (or “distractor” stimulus) should have a greater impact or disruption on timing than a fixed one. The second prediction is that when comparing two intervals, timing of the second will be influenced by the first if they are not separated by more time than the reset threshold. These predictions can be tested psychophysically to distinguish the extent to which various types of timing arise from network clocks. In the auditory system, multiple studies examining interval discrimination with reset tasks have shown results consistent with an SDN model (Karmarkar and Buonomano, 2007; Buonomano et al., 2009; Spencer et al., 2009). Based on this data, the influence of the SDN appears to fall off somewhere between intervals above 250 but below 500 ms (Buonomano et al., 2009). This limit is potentially as restrictive as 300 ms, since stimuli including that interval do not show the expected pattern of impairment due to a variable context (Spencer et al., 2009). However, a secondary analysis of the data from that study might suggest that distractor signals prior to the target interval do appear to exert some bias on discrimination as long as the whole stimulus sequence, or temporal object, does not significantly exceed 400 ms (Figures 3 and 5, Spencer et al., 2009). A threshold in this vicinity is consistent with the assumptions of the SDN model since it is dependent on the time constants of short-term plasticity, which are on the order of a few hundreds of milliseconds (e.g. Zucker, 1989). Such a restriction in range could be perceived as a challenge to the relevance of SDNs, or network clocks in general, to other sensory modalities. Though intervals of less than 300–400 ms can be useful for somatosensory timing, the visual system appears to operate on a fundamentally slower timescale. Experiments in which individuals had to reproduce durations represented by visual stimuli showed that participants’ lowest estimated duration was 300 ms, even when the target interval was less than 100 ms, suggesting difficulty in accurately perceiving those shorter times (Lewis and Miall, 2009). In addition, visual discrimination of intervals has been shown to be less precise, that is, to have a higher variance, than auditory perception of the same durations (Merchant et al., 2008), which could prevent effective measurement of time spans less than 200 ms. Based on this data, the range over which the SDN operates for auditory stimuli might appear to render it inconsequential for visual ones. It should be noted though, that previous studies have revealed some visual discriminatory capabilities for intervals in the 100 to 200-ms range (e.g. Mattes and Ulrich, 1998; Westheimer, 1999). Furthermore, interval timing in the visual system is spatially localized (Johnston et al., 2006; Burr et al., 2007), suggesting that the mechanism is specific to early visual cortices, consistent with the idea of a local network clock. As such, it is possible that some of the reduction in precision for interval timing is due to the inherent variance of the basic response time in primary visual cortex. This could be considered a time-independent issue that influences the system by adding noise rather than indicating a non-SDN mechanism. Vision has been considered a difficult modality for defining or studying temporal processing mechanisms. This is because visual timing is extremely sensitive to a number of atemporal stimulus characteristics (Eagleman, 2008), and is generally tightly linked to spatial information, as in motion detection. However, we propose that it is key in determining whether the SDN can be considered a general model of sensory timing. Finally, it is important to recognize that there are other network clock models based on differing circuitry (e.g. Buonomano and Mauk, 1994; Medina et al., 2000; Fiete et al., 2010). This reflects the broader concept of an intrinsic timer, that temporal processing is dependent on the specific properties of the neural locus in question. The differences in the structure of these models also lead to differences in the range of times they can process, and their ability to integrate spatial and temporal information. For example, it has been proposed that population clock models that leverage recurrent excitatory connections with strong weights can account for motor timing that extends from milliseconds into seconds (Buonomano and Laje, 2010). Despite this diversity, network clock models largely show the same phenotype of continuous temporal processing, in which sequences are treated as temporal objects, making reset-type tasks a general diagnostic tool for this class of timers. As a result, using psychophysical measures to investigate timing directly across modalities is an important first step in defining the contribution of network clock models to human interval discrimination.