Neurally-constrained modeling of human gaze strategies in a change blindness task

Despite possessing the capacity for selective attention, we often fail to notice the obvious. We investigated participants’ (n = 39) failures to detect salient changes in a change blindness experiment. Surprisingly, change detection success varied by over two-fold across participants. These variations could not be readily explained by differences in scan paths or fixated visual features. Yet, two simple gaze metrics–mean duration of fixations and the variance of saccade amplitudes–systematically predicted change detection success. We explored the mechanistic underpinnings of these results with a neurally-constrained model based on the Bayesian framework of sequential probability ratio testing, with a posterior odds-ratio rule for shifting gaze. The model’s gaze strategies and success rates closely mimicked human data. Moreover, the model outperformed a state-of-the-art deep neural network (DeepGaze II) with predicting human gaze patterns in this change blindness task. Our mechanistic model reveals putative rational observer search strategies for change detection during change blindness, with critical real-world implications.     

An implicit measure of undetected change

Several paradigms (e.g. change blindness, inattentional blindness, transsaccadic integration) indicate that observers are often very poor at reporting changes to their visual environment. Such evidence has been used to suggest that the spatio-temporal coherence needed to represent change can only occur in the presence of focused attention. However, those studies almost always rely on explicit reports. It remains a possibility that the visual system can implicitly detect change, but that in the absence of focused attention, the change does not reach awareness and consequently is not reported. To test this possibility, we used a simple change detection paradigm coupled with a speeded orientation discrimination task. Even when observers reported being unaware of a change in an item's orientation, its final orientation effectively biased their response in the orientation discrimination task. Both in aware and unaware trials, errors were most frequent when the changed item and the probe had incongruent orientations. These results demonstrate that the nature of the change can be represented in the absence of awareness.     

Attentive and pre-attentive processes in change detection and identification

In studies of change blindness, observers often have the phenomenological impression that the blindness is overcome all at once, so that change detection, localization and identification apparently occur together. Three experiments are described that explore dissociations between these processes using a discrete trial procedure in which 2 visual frames are presented sequentially with no intervening inter-frame-interval. The results reveal that change detection and localization are essentially perfect under these conditions regardless of the number of elements in the display, which is consistent with the idea that change detection and localization are mediated by pre-attentive parallel processes.In contrast, identification accuracy for an item before it changes is generally poor, and is heavily dependent on the number of items displayed. Identification accuracy after a change is substantially better, but depends on the new item's duration. This suggests that the change captures attention, which substantially enhances the likelihood of correctly identifying the new item. However, the results also reveal a limited capacity to identify unattended items. Specifically, we provide evidence that strongly suggests that, at least under these conditions, observers were able to identify two items without focused attention. Our results further suggest that spatial pre-cues that attract attention to an item before the change occurs simply ensure that the cued item is one of the two whose identity is encoded.     

Comparison of identified mitral and tufted cells in freely breathing rats: II. Odor-evoked responses

Mitral and tufted cells are the 2 types of output neurons of the main olfactory bulb. They are located in distinct layers, have distinct projection patterns of their dendrites and axons, and likely have distinct relationships with the intrabulbar inhibitory circuits. They could thus be functionally distinct and process different aspects of olfactory information. To examine this possibility, we compared the odor-evoked responses of identified single units recorded in the mitral cell layer (MCL units), in the core of the external plexiform layer (not at the glomerular border tufted cells), or at the glomerular border of this layer (GB tufted cells) of the entire olfactory bulb. Differences between mitral and tufted cells were observed only when subtle aspects of the responses were explored, such as the firing rate per respiratory cycle or the distribution of firing activity along the respiratory cycle. By contrast, more clear differences were found when the 2 subtypes of tufted cells were examined separately. GB units were significantly more responsive, had significantly higher firing activity, and showed greater activity at the transition between inspiration and expiration. The projection-type tufted cells situated closer to the entrance of the olfactory bulb may thus form a distinct physiological class of output neurons and differ from mitral cells and other tufted cells in the manner of processing olfactory information.     

Working Memory Cells' Behavior May Be Explained by Cross-Regional Networks with Synaptic Facilitation

Neurons in the cortex exhibit a number of patterns that correlate with working memory. Specifically, averaged across trials of working memory tasks, neurons exhibit different firing rate patterns during the delay of those tasks. These patterns include: 1) persistent fixed-frequency elevated rates above baseline, 2) elevated rates that decay throughout the tasks memory period, 3) rates that accelerate throughout the delay, and 4) patterns of inhibited firing (below baseline) analogous to each of the preceding excitatory patterns. Persistent elevated rate patterns are believed to be the neural correlate of working memory retention and preparation for execution of behavioral/motor responses as required in working memory tasks. Models have proposed that such activity corresponds to stable attractors in cortical neural networks with fixed synaptic weights. However, the variability in patterned behavior and the firing statistics of real neurons across the entire range of those behaviors across and within trials of working memory tasks are typical not reproduced. Here we examine the effect of dynamic synapses and network architectures with multiple cortical areas on the states and dynamics of working memory networks. The analysis indicates that the multiple pattern types exhibited by cells in working memory networks are inherent in networks with dynamic synapses, and that the variability and firing statistics in such networks with distributed architectures agree with that observed in the cortex.     

A New Approach for Determining Phase Response Curves Reveals that Purkinje Cells Can Act as Perfect Integrators

Cerebellar Purkinje cells display complex intrinsic dynamics. They fire spontaneously, exhibit bistability, and via mutual network interactions are involved in the generation of high frequency oscillations and travelling waves of activity. To probe the dynamical properties of Purkinje cells we measured their phase response curves (PRCs). PRCs quantify the change in spike phase caused by a stimulus as a function of its temporal position within the interspike interval, and are widely used to predict neuronal responses to more complex stimulus patterns. Significant variability in the interspike interval during spontaneous firing can lead to PRCs with a low signal-to-noise ratio, requiring averaging over thousands of trials. We show using electrophysiological experiments and simulations that the PRC calculated in the traditional way by sampling the interspike interval with brief current pulses is biased. We introduce a corrected approach for calculating PRCs which eliminates this bias. Using our new approach, we show that Purkinje cell PRCs change qualitatively depending on the firing frequency of the cell. At high firing rates, Purkinje cells exhibit single-peaked, or monophasic PRCs. Surprisingly, at low firing rates, Purkinje cell PRCs are largely independent of phase, resembling PRCs of ideal non-leaky integrate-and-fire neurons. These results indicate that Purkinje cells can act as perfect integrators at low firing rates, and that the integration mode of Purkinje cells depends on their firing rate.     

Adaptation supports short-term memory in a visual change detection task

The maintenance of short-term memories is critical for survival in a dynamically changing world. Previous studies suggest that this memory can be stored in the form of persistent neural activity or using a synaptic mechanism, such as with short-term plasticity. Here, we compare the predictions of these two mechanisms to neural and behavioral measurements in a visual change detection task. Mice were trained to respond to changes in a repeated sequence of natural images while neural activity was recorded using two-photon calcium imaging. We also trained two types of artificial neural networks on the same change detection task as the mice. Following fixed pre-processing using a pretrained convolutional neural network, either a recurrent neural network (RNN) or a feedforward neural network with short-term synaptic depression (STPNet) was trained to the same level of performance as the mice. While both networks are able to learn the task, the STPNet model contains units whose activity are more similar to the in vivo data and produces errors which are more similar to the mice. When images are omitted, an unexpected perturbation which was absent during training, mice often do not respond to the omission but are more likely to respond to the subsequent image. Unlike the RNN model, STPNet produces a similar pattern of behavior. These results suggest that simple neural adaptation mechanisms may serve as an important bottom-up memory signal in this task, which can be used by downstream areas in the decision-making process.     

Single unit activity in the mesencephalon of Sternarchus

Action potentials evoked in a phase-locked 1 : 1 relationship by natural electric organ discharges (EOD) were recorded extracellularly and intracellularly from single mesencephalic magnocellular nucleus units in the high frequency electric fish Sternarchus albifrons (Gymnotidae). This activity has been shown to be the result of an extrinsic feedback of the electrosensory system and is probably important for the socalled jamming avoidance response triggered artificial electric pulses when delivered into the water in a 1 : 1 relationship at intensities higher than the EOD. In the same way, artificial pulses of frequency near EOD could either drive or, due to beats, greatly disturb the regular firing of the units. More insight into the neural mechanism was yielded by single EOD-triggered shocks provoking a failure in firing of certain action potentials of the series and causing long-lasting (10-20 ms) accelerations and decelerations of the regular EOD-evoked firing (transient disturbance). Intracellular stimulations show similar effects. The biological significance of this neural mechanism for the fish's electroperception and JAR is discussed.     

Simultaneous single unit recording in the mitral cell layer of the rat olfactory bulb under nasal and tracheal breathing

Odor perception depends on the odorant-evoked changes on Mitral/Tufted cell firing pattern within the olfactory bulb (OB). The OB exhibits a significant "ongoing" or spontaneous activity in the absence of sensory stimulation. We characterized this ongoing activity by simultaneously recording several single neurons in the mitral cell layer (MCL) of anesthetized rats and determined the extent of synchrony and oscillations under nasal and tracheal breathing. We recorded 115 neurons and found no significant differences in the mean firing rates between both breathing conditions. Surprisingly, nearly all single units exhibited a long refractory period averaging 14.4 ms during nasal respiration that was not different under tracheal breathing. We found a small incidence (2% of neurons) of gamma band oscillations and a low incidence (8.1%) of correlated firing between adjacent MCL cells. During nasal respiration, a significant oscillation at the respiratory rate was observed in 12% of cells that disappeared during tracheal breathing. Thus, in the absence of odorants, MCL cells exhibit a long refractory period, probably reflecting the intrinsic OB network properties. Furthermore, in the absence of sensory stimulation, MCL cell discharge does not oscillate in the gamma band and the respiratory cycle can modulate the firing of these cells.     

The dark side of visual attention

The limited capacity of neural processing restricts the number of objects and locations that can be attended to. Selected events are readily enhanced: the bright side of attention. However, such focal processing comes at a cost, namely, functional blindness for unattended events: the dark side of visual attention. Recent work has advanced our understanding of the neural mechanisms that facilitate visual processing, as well as the neural correlates of unattended, unconscious visual events. Also, new results have revealed how attentional deployment is optimized by non-visual factors such as behavioral set, past experience, and emotional salience.     

Implicit Binding of Facial Features During Change Blindness

Change blindness refers to the inability to detect visual changes if introduced together with an eye-movement, blink, flash of light, or with distracting stimuli. Evidence of implicit detection of changed visual features during change blindness has been reported in a number of studies using both behavioral and neurophysiological measurements. However, it is not known whether implicit detection occurs only at the level of single features or whether complex organizations of features can be implicitly detected as well. We tested this in adult humans using intact and scrambled versions of schematic faces as stimuli in a change blindness paradigm while recording event-related potentials (ERPs). An enlargement of the face-sensitive N170 ERP component was observed at the right temporal electrode site to changes from scrambled to intact faces, even if the participants were not consciously able to report such changes (change blindness). Similarly, the disintegration of an intact face to scrambled features resulted in attenuated N170 responses during change blindness. Other ERP deflections were modulated by changes, but unlike the N170 component, they were indifferent to the direction of the change. The bidirectional modulation of the N170 component during change blindness suggests that implicit change detection can also occur at the level of complex features in the case of facial stimuli.     

Single motor unit activity in the diaphragm of cat during pressure breathing

In this study we analyzed the breath-by-breath activity of single motor units in the diaphragm slip of allobarbital-anesthetized cats during quiet breathing and during continuous positive- and negative-pressure breathing. Our objective was to determine whether single motor units, on the basis of their activities, can be separated into discrete subpopulations or whether they fall on a continuum analogous to that of motor units of hindlimb muscles. The firing profiles of each unit were characterized for each pressure level by the onset and peak firing frequencies, onset latency, duration of firing, number of impulses per breath, and minimal frequency, when appropriate. Units with shorter onset latencies had higher peak frequencies, longer firing durations, and increased firing frequencies than did units with longer onset latencies. These comparative relationships persisted even though the activity of every motor unit was altered during pressure breathing. During positive-pressure breathing onset latencies were lengthened, and durations of firing were shortened with little change in onset or peak frequencies. Late units might be silenced. During negative-pressure breathing onset latencies were shortened, and durations of firing were lengthened, sufficiently in some cases to fill the expiratory pause. In addition, previously inactive units were recruited late in inspiration for short, relatively high frequency bursts during inspiration. The results support the concept that the phrenic motoneuron pool is comprised of three discrete subpopulations.     

Electrophysiological correlates of synchronous neural activity and attention: a short review

Attentional selection implies preferential treatment of some sensory stimuli over others. This requires differential representation of attended and unattended stimuli. Most previous research has focused on pure rate codes for this representation but recent evidence indicates that a mixed code, involving both mean firing rate and temporal codes, may be employed. Of particular interest is a distinction of attended from unattended stimuli based on synchrony within neural populations. I review electrophysiological evidence at macroscopic, mesoscopic and microscopic spatial scales showing that the degree of synchronous activity varies with the attentional state of the perceiving organism.     

Mass correlograms of multiple neuronal activity in the cat's extrastriate cortex

Electrical activity of a population of visually responsive cells located in the vicinity of a single functionally defined neuron was recorded in the area 18 of the cat's cerebral cortex with a single tungsten microelectrode. The correlograms calculated from the mass activity record showed an existence of a rhythmic neuronal firing with an average interval near to 3 ms. When the system was activated by a visual stimulus, a line at an optimal angle moving in an optimal direction, the rhythmic activity became regular, acquiring an oscillatory sinusoidal character. This rhythmic pattern cannot be easily recognized when the activity of a single neuron is recorded. It is possible that such rhythmic activity involving large numbers of neurons contributes to the recognition of the velocity and position of the visual stimulus.     

Tactile-spatial and cross-modal attention effects in the second somatosensory and 7b cortical areas of rhesus monkeys

Abstract This study analyzed neuronal responses in the second somatosensory (SII) and 7b cortical areas during a selective attention task. Cues directed attention to one of three simultaneous stimuli: vibrotactile stimuli applied to mirror sites on both hands or to a similarly timed auditory tone. Two stimulus patterns appeared with equal probability for the cued stimulus: a constant amplitude sinewave or the latter with a superimposed brief amplitude pulse midway in the trial. Uncued stimuli always contained amplitude pulses. Monkeys demonstrated whether an amplitude pulse at the cued location was present or absent by making appropriately rewarded up and down foot pedal movements. Cue location and stimulus pattern varied trial-wise and pseudo-randomly. Average firing rates to vibrotactile stimuli in 82 of 181 SII cells and 13 of 22 area 7b cells differed significantly during at least one epoch for trials cued to the contralateral hand when compared to trials cued to the ipsilateral hand or auditory stimulus. Predominant were relatively suppressed firing rates during times prior to the epoch containing the amplitude pulses or enhanced activity during and after these pulses. Generally, different cells showed suppression early vs enhancement later in a trial. Analyses of the ratio between firing rates before and during the amplitude pulses suggested improved evoked signals to the amplitude pulses. The discussion considers attention as a mechanism for reducing distractions, early in the trial through suppressing these signals, or for selectively increasing response magnitudes in the cued channel, especially around times when amplitude pulses were present or absent.     

Ia-afferent input to motoneurons during shortening and lengthening muscle contractions in humans.

The central nervous system employs different strategies to execute specific motor tasks. Because afferent feedback during shortening and lengthening muscle contractions differs, the neural strategy underlying these tasks may be quite distinct. Cortical drive may be adjusted or afferent input regulated. The exact mechanisms are not clear. Here, we examine the control of synaptic transmission across the Ia synapse during shortening and lengthening muscle contractions. Subjects were instructed to maintain isolated activity in a single tibialis anterior (TA) motor unit while muscle length was varied from flexion to extension and back. At a fixed interval after a firing of the active motor unit, a single electrical stimulus was applied to the common peroneal nerve to activate Ia afferents from the TA muscle. We investigated the stimulus-induced change in firing probability of 19 individual low-threshold TA motor units during shortening and lengthening contractions. Any change in firing probability depends on both pre- and postsynaptic mechanisms. In this experiment, motoneuron firing rate was similar during both contraction types. There was no difference in the firing probability between shortening and lengthening contractions (0.23 +/- 0.03 and 0.20 +/- 0.02, respectively). We suggest that there is no contraction type-specific control of Ia input to the motoneurons during shortening and lengthening muscle contractions. Cortical adjustments may have occurred.     

The functional role of different neural activation profiles during precision grip: An artificial neural network approach

A dynamic and recurrent artificial neural network was used to investigate the functional properties of firing patterns observed in the primary motor (M1) and the primary somatosensory (S1) cortex of the behaving monkey during control of precision grip force. In the behaving monkey it was found that neurons in M1 and in S1 increase their firing activity with increasing grip force, as do the intrinsic and extrinsic hand muscles implicated in the task. However, some neurons also decreased their activity as a function of increasing force. The functional implication of these latter neurons is not clear and has not been elucidated so far. In order to explore their functional implication, we therefore simulated patterns of neural activity in artificial neural networks that represent cortical, spinal and afferent neural populations and tested whether particular activity profiles would emerge as a function of the input and of the connectivity of these networks. The functional implication of units with emergent or imposed decreasing activity was then explored.Decreasing patterns of activity in M1 units did not emerge from the networks. However, the same networks generated decreasing activity if imposed as target patterns. As indicated by the emerging weight space, M1 projection units with decreasing patterns are functionally less involved in driving alpha motoneurons than units with increasing profiles. Furthermore, these units did not provide significant fusimotor drive, whereas those with increasing profiles did. Fusimotor drive was a function of the (imposed) form of muscle spindle afferent activity: with gamma (fusimotor) drive, muscle spindle afferents provided signals other than muscle length (as observed experimentally). The network solutions thus predict a functional dichotomy between increasing and decreasing M1 neurons: the former primarily drive alpha and gamma motoneurons, the latter only weakly alpha motoneurons.     

Lucid dreaming frequency and change blindness performance.

There are reports of lucid dreaming being cued by the recognition that a dream event is bizarre from the point of view of waking life. However, for dreams in general, there is a lack of ability to notice or question bizarre occurrences. A waking-life analog of this inability is here proposed to be change blindness. In change blindness tasks, a prominent alteration to a photograph occurs repeatedly, but it is rare for these changes to be spotted immediately. It was hypothesized that lucid dreamers would perform better on change blindness tasks than would nonlucid dreamers. Contrary to the hypothesis, individuals who reported having lucid dreams more than once per month (n = 13), occasional lucid dreamers (n = 13), and nonlucid dreamers (n = 12) were found not to differ significantly on performance on 6 change blindness tasks. How the usually proficient unconscious detection of errors during waking life is disabled during dreams remains to be determined, but it does not seem from the results here to have a simple relationship with the waking-life phenomenon of change blindness. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Patterned response to odor in single neurones of goldfish olfactory bulb: influence of odor quality and other stimulus parameters

Responses of 75 single units in the goldfish olfactory bulb were analyzed in detail for their relationship to the time-course of the change in odor concentration during each odor stimulus. Odor stimuli were controlled for rise time, duration, and peak concentration by an apparatus developed for the purpose. This apparatus enabled aqueous odor stimuli to be interposed into a constant water stream without changes in flow rate. The time-course of the concentration change within the olfactory sac was inferred from conductivity measurements at the incurrent and excurrent nostrils. Temporal patterns of firing rate elicited by stimuli with relatively slow rising and falling phases could be quite complex combinations of excitation and suppression. Different temporal patterns were produced by different substances at a single concentration in most units. Statistical measures of the temporal pattern of response for a small number of cells at a given concentration were more characteristic of the stimulus substance than any of three measures of magnitude of response. The temporal patterns change when the peak concentration, duration, and rise time of the stimuli are varied. The nature of these changes suggests that the different patterns are due primarily to the combined influence of two factors: (a) a stimulus whose concentration varies over time and (b) a relationship between concentration and impulse frequency which varies from unit to unit. Some units produce patterns suggestive of influence by neural events of long time constant. The importance of temporal patterns in odor quality and odor intensity coding is discussed.     

Membrane Potential Dynamics of Spontaneous and Visually Evoked Gamma Activity in V1 of Awake Mice

Cortical gamma activity (30–80 Hz) is believed to play important functions in neural computation and arises from the interplay of parvalbumin-expressing interneurons (PV) and pyramidal cells (PYRs). However, the subthreshold dynamics underlying its emergence in the cortex of awake animals remain unclear. Here, we characterized the intracellular dynamics of PVs and PYRs during spontaneous and visually evoked gamma activity in layers 2/3 of V1 of awake mice using targeted patch-clamp recordings and synchronous local field potentials (LFPs). Strong gamma activity patterned in short bouts (one to three cycles), occurred when PVs and PYRs were depolarizing and entrained their membrane potential dynamics regardless of the presence of visual stimulation. PV firing phase locked unconditionally to gamma activity. However, PYRs only phase locked to visually evoked gamma bouts. Taken together, our results indicate that gamma activity corresponds to short pulses of correlated background synaptic activity synchronizing the output of cortical neurons depending on external sensory drive.