The effect of visual long-term memory on change detection

Megumi Nishiyama and Jun Kawaguchi

According to recent studies, relatively detailed visual representations are formed instantly and retained long. However, the relationships between the retained long-term memories and current visual perception and cognition have not been revealed. Therefore, we aimed to investigate the influence of visual long-term memory on change detection accuracy. The experiment consisted of three phases. In a study phase, images containing six meaningless objects were serially presented while participants performed a repetition-detection task. Then, in a change-detection phase, on each trial, pre- and post-change images were presented once with a short interval. Half of the pre-change images were identical to the study phase (studied condition), while the other half were novel (novel condition). In a recognition phase, the images in the study phase were presented and tested to confirm whether they were retained as visual long-term memory. It was revealed that the change detection was more accurate in the studied condition than in the novel condition, and the visual memory formed in the study phase could be retained after the change detection task. It was suggested that visual long-term memory is robust and the memory is available for use in current visual perception.

Social influence alters perception

Jacob Jolij

In one of the classic studies in psychology, Asch demonstrated that social pressure may lead people to give obviously wrong answers in a perceptual decision task. This effect is typically attributed to a deliberate strategy - people do not like to stand out, and conform to a group response, even though their perception tells them they're wrong. However, is this view correct? In this talk, I will present some data that suggests that social pressure may actually change people's perceptions, as measured with evoked brain potentials. In a visual detection task, we let participants detect faces whilst they were influenced by two conferderates. We measured EEG to assess whether any changes in responses were accompanied by changes in perception-related brain activity. We managed to replicate the classic Asch-effect, that is, participants conformed their responses to the confederates. Interestingly, these changes in responses were accompanied by changes in brain activity: false alarms were accompanied by perception-related visual evoked potentials (in particular the N200). This result shows that we do not only see what's out there, or even what we expect or feel - to some extent, we also see what others tell us to see.

Neural activity in human visual cortex predicts bistable perceptual state durations

Niels Kloosterman, Arjan Hillebrand, Bob van Dijk, Victor Lamme and Tobias Donner

During motion-induced blindness (MIB), a salient target stimulus disappears from perception for variable periods of time when surrounded by a moving pattern. We recorded the magneto-encephalogram (MEG) to characterize the temporal fine structure of neural activity in visual cortex during MIB. In different sessions, subjects reported target disappearance by either pressing or releasing a button. MEG power over visual cortex in the 9-30 Hz frequency range decreased transiently around target disappearance reports, followed by a sustained suppression within the < 9 Hz and 9-30 Hz ranges. The sign of these power modulations reflected the perceptual reports (target disappearance/reappearance), but not the motor responses (button press/release). The modulations were not confined to the cortical sub-regions processing the target: they were equally strong over both hemispheres, irrespective of the target occurring in one visual hemifield. The same modulations occurred when subjects reported on the physical removal of the target ('replay' condition). Remarkably, the amplitude of the transient modulation predicted the duration of the subsequent disappearance state during MIB, but not replay. We conclude that the modulations reflect a state change throughout visual cortex associated with the perceptual decision, which, in turn, stabilizes the subsequent bistable perceptual state.

Sensory processing in the absence of conscious awareness improves decision-making

Alexandra Vlassova and Joel Pearson

Recently there has been a surge of research aimed at discovering the role of conscious awareness in the context of decision-making. The controversial claim that information can be processed and evaluated in the absence of conscious awareness has been heavily debated in the literature. We address this dispute through a novel paradigm that allows us to control and manipulate both awareness and the decision variables. By presenting dynamic noise patterns binocularly, we were able to suppress random-dot-motion (RDM) stimuli from conscious awareness. RDM stimuli require the gradual accumulation of evidence over time, and therefore provide a useful analogue for the processes underlying naturalistic decision-making. We found that information that was presented outside of awareness was accumulated and used to improve decision accuracy. This improvement was not accompanied by a similar boost to confidence. These results indicate that perceptual and metacognitive awareness is not necessary for evidence to be accumulated and an accurate decision to be made.

Relation between electrophysiological correlates of affective conditioning and the discriminability and detectability of stimuli in metacontrast masking

Philipp Hintze, Markus Junghöfer and Maximilian Bruchmann

We introduce a new paradigm combining affective conditioning and metacontrast masking. We study the interplay of the affective value of a stimulus and its visibility by manipulating the latter in several consecutive steps. One of two grating stimuli is combined with an aversive auditory startle in a trace conditioning paradigm. This grating thereby acquires a negative emotional valence while the other remains neutral. Before, during and after conditioning, we obtain metacontrast masking functions of the two gratings by varying the stimulus onset asynchrony (SOA) between target and mask. We observe a strong dissociation between detection and discrimination performance: At short SOAs, subjects reliably perceive the presence of a target stimulus yet fail to identify it. With increasing SOA, the discrimination improves while detection performance deteriorates. These effects were unaffected by conditioning. In a study combining behavioral measurements with simultaneous MEG and EEG recordings, we identify the neural sources of affective perception by means of a current density reconstruction. We thereby determine its dependence on stimulus detection versus stimulus identification.

Unconscious evaluative conditioning: Attitude change induced by invisible words

Su-Ling Yeh, Hung-Yu Lin, Shuo-Heng Li and Meng-Ning Tsai

To what extent can unconscious processing affect the choices of individuals ? We have shown previously that invisible words can be processed up to the semantic level. Here, we further demonstrate that through evaluative conditioning, invisible words can also induce attitude change to neutral stimuli. A word (US) and a flag (CS) are paired and presented to one eye, but made invisible by continuous flash suppression to the other eye. The word used was 'like' or 'dislike', and the flag chosen from a set of flags given neutral valence in a pre-test. Superimposed on the visible masks were two horizontal lines within which the invisible word-flag pair was embedded. Participants judged the relative length of the two lines and verified the invisibility of the word-flag pair after each trial. After 12 trials of each of 8 US-CS pairing, participants gave their emotional evaluations for each flag. Flags that had been paired with the word 'like' were rated as positive valence in the post-test; however, this effect did not exist in the visible (binocularly viewing) condition. The fact that evaluative conditioning occurs for unconsciously (but not consciously) perceived word-flag pairs uncovers a new facet of the nature of consciousness.

Amygdala response for faces in a patient with cortical blindness

Alan Pegna, Alexis Hervais-Adelman, Legrand Lore, Marzia Del Zotto and Beatrice de Gelder

Bilateral destruction of primary visual areas leads to blindness of cortical origin. Here, we report the case of a well-known cortically-blind patient presents affective blindsight. Previous observations had shown that emotional facial expressions gave rise to right amygdala activation. In this follow-up study, we carried out a block-design fMRI procedure in which neutral faces were presented along with other objects (cars, bodies, butterflies and scrambled figures) that could not be reported by the patient. When faces were contrasted with other stimuli both amygdalae were found to be activated. Affective blindsight therefore seems to extend beyond emotional stimuli to include behaviourally-relevant stimuli on a more global scale.

Dissociation between eye movements and motion perception in bistable motion plaid perception for 1D motion, not 2D motion

Jean-Michel Hupé

Usually eye movements are consistent with the direction of perceived motion. Recently however, it was shown that reflexive eye movements followed the vector average of orthogonal moving gratings presented dichoptically, while subjects perceived only one grating and motion direction [Spering et al, 2011, Psychological Science, 22, 216-225]. When both gratings are presented to both eyes, stimuli are ambiguous plaids that can be perceived either as a coherent pattern moving rigidly or as two gratings sliding over each other. Prolonged observation leads to bistable alternation between coherence and transparency. Even under steady fixation moving plaids can generate small optokinetic nystagmus (OKN) along the directions of both component (1D) and pattern (2D) motion. 1D OKN followed the direction of either one or the other grating, not the vector average (whose direction was the same as 2D motion). 2D OKN were twice more frequent when coherence was reported than when transparency was reported, while 1D OKN were not modulated by perception. These results confirm the possible dissociation between eye movements and motion perception with stimuli presented over long durations, but only for 1D motion, not 2D motion. Supported by Agence Nationale de Recherche ANR-08-BLAN-0167-01

Co-ordination of voluntary and automatic eye-movements: Accuracy of saccades made during concomitant optokinetic nystagmus

James Harrison, Petroc Sumner and Tom Freeman

During self-motion, saccades intended to foveate targets of interest must co-occur with eye movements which stabilize the retinal image, such as optokinetic nystagmus (OKN) and the vestibular-ocular reflex (VOR). Despite the close relationship between target-selection and gaze-stabilizing eye movements, they are typically regarded as being fundamentally different, with a sharp distinction drawn between conscious and unconscious motor actions. So how can the voluntary saccadic system effectively accommodate simultaneous reflexive nystagmus? We examined this question by asking observers to make saccades to vertically-flashed targets while executing horizontal OKN. Results showed that saccades were only partially compensated for displacements of the eye due to OKN, with further experiments showing that compensation was improved when the displacement was due to voluntary smooth-pursuit. Furthermore, analysis of the saccade trajectories showed that they became curved in a way consistent with the idea that saccades may be influenced by activation of the OKN system. These findings have implications for our understanding of how voluntary and automatic movements are co-ordinated in the human brain, as well as our understanding of the difficulties an active observer faces in orienting to targets during concomitant self-motion.

Delusions and the role of beliefs in perceptual inference

Philipp Sterzer, Marcus Rothkirch, Maria Sekutowicz, Hannes Rössler, John-Dylan Haynes and Katharina Schmack

Delusions are unfounded yet tenacious beliefs and a symptom of psychotic disorder. Varying degrees of delusional ideation are also found in the healthy population. Here, we experimentally tested a neurocognitive model that explains the formation and persistence of delusional beliefs in terms of altered perceptual inference. We performed two behavioral experiments and a functional magnetic resonance imaging experiment in healthy participants to probe the relationship between delusion-proneness and the effect of learned predictions on bistable visual perception using an ambiguous random-dot kinematogram. Behaviorally, delusional ideation was associated with less perceptual stability during intermittent viewing of the ambiguous stimulus but a stronger influence of experimentally induced beliefs on perception. The perceptual effect of beliefs was reflected in fMRI signal patterns in visual cortex, from which biased perception could be decoded using multivoxel pattern analysis. Furthermore, delusional ideation was associated with enhanced functional connectivity between an orbitofrontal region that encoded beliefs and visual cortex. Our results indicate that beliefs can influence the contents of awareness through feedback signals from higher-order brain areas to sensory cortices. In delusion-prone individuals, an enhanced effect of beliefs on sensory processing may sculpt perception into conformity with beliefs, thereby contributing to the tenacious persistence of delusional beliefs.

What's so special about motion? Its' all in the tail!

Derek Arnold, Welber Marinovic and David Whitney

It has been suggested that there is a perceptually explicit predictive signal at the leading edges of movement that summates with low contrast targets. Evidence for this is that sensitivity for low contrast target waveforms at leading edges of high contrast inducing gratings varies with the phasic relationship between the abutting stimuli, but this does not happen at trailing edges of movement [1]. While this has been attributed to a predictive signal unique to the leading edges of movement, an alternate possibility is that summation simply involves the abutting waveforms, but this summation is selectively inhibited at trailing edges of motion. We assessed this possibility by examining target sensitivity at leading and trailing edges of motion, and by comparing these measures with those for flickering waveforms. Sensitivity for abutting flickers qualitatively matched sensitivity at leading edges of motion; both were marked by phase dependent summation. Thus what differentiates movement from directionless flicker in this context is that sensitivity at trailing edges displays no evidence of summation. We attribute this to an inhibitory operation, involved in suppressing blur signals that usually result from retinal motion. 1. Roach McGraw & Johnston (2011) Visual motion induces a forward prediction of spatial pattern. Current Biology.

Timing the Rubber Hand Illusion: Owning the rubber hand involves substitution, not addition

An-Yi Chang, Timothy Lane and Su-Ling Yeh

We assume biological hands connected to our bodies belong to self; objects not so connected do not. In the Rubber Hand Illusion (RHI), however, observing a rubber hand being stroked while the occluded real hand is synchronously stroked engenders a feeling that the rubber hand belongs to self. Our investigation had two foci: assessment of RHI's temporal dimension and of competing hypotheses. The former was motivated by dissatisfaction with other behavioral proxies; the latter, intended to determine whether experiencing a rubber hand as mine (ownership) involves the sense that my real hand has been replaced (disownership), or the sense that one has acquired a supernumerary hand. Participants indicated elapsed time for onset of ownership and disownership by stepping on a pedal. Also, after each trial, a questionnaire containing two sets of items-ownership and disownership (derived from Longo et al., 2008, Cognition, 107, 978-998)-was administered. By both measures the experience of ownership is revealed to be stronger than that of disownership; however, results also indicate that the stronger the ownership experience, the stronger is the experience of disownership. In short, timing RHI and assessment of the relationship between ownership and disownership suggest that RHI involves substitution, not addition.

Hearing lips and (not) seeing voices: Audiovisual integration with the suppressed percept

Manuel Vidal and Victor Barrès

In binocular rivalry, sensory input remains the same yet subjective experience fluctuates irremediably between two mutually exclusive representations. We investigated whether a suppressed viseme can still produce the McGurk effect with an auditory input, which would indicate that unconscious lips motion sometimes modulate the perceived phoneme through the well-known multisensory integration mechanism. We used speech stimuli for it involves robust audiovisual interactions at several cortical levels. The perceived phoneme when presenting a synchronous voice saying /aba/ together with rivaling faces saying /aba/ and /aga/ was recorded for the seven McGurk sensitive participants. We found that when the dominant percept was seen with the non-dominant eye, in about 20% of the trials the audiovisual outcome resulted from an integration with the suppressed percept. This integration could either produce or cancel the McGurk effect that was expected with the actually seen viseme. This observation raises serious questions in the fields of speech perception and multisensory binding, suggesting that feature binding might not be a prerequisite to perceptual awareness. Further experiments are being conducted to determine whether the information binding failed within the same modality (color and lips motion) or between the two modalities (lips motion and voice).

The influence of rotational speed and line width on perceptual reversals of the Lissajous figure

Veith Weilnhammer, Karin T. Ludwig Ludwig, Philipp Sterzer and Guido Hesselmann

Two superimposed and increasingly phase-shifted sinusoidal functions can induce the percept of an object spontaneously changing its direction of rotation, dubbed the Lissajous figure. Here, we investigated the influence of rotational speed and line width on the perceptual reversals of bistable Lissajous figures. By correcting subjects'response latencies with reaction times from a disambiguated replay condition, we found that perceptual switches occurred most frequently when the waveforms overlapped, i.e., at depth symmetric displays. At those critical positions, reversal probabilities linearly increased with speed and line width, which ranged from 2 to 9 seconds per rotation and 0.02° to 0.2° of visual angle, respectively. Thus, switch probabilities were highest for slowly rotating stimuli. Interestingly, switches also tended to occur more often for thicker lines. Perceptual switches of the Lissajous figure demand a reassignment of points from background to foreground, and vice versa, which might be facilitated at depth symmetric stimulus configurations. For slowly turning and thick lined Lissajous patterns, these overlaps last longer, thus increasing switch probability. Our findings fit nicely with the hypothesis that prior experience of the visual environment prevents observers from randomly switching to the otherwise equally plausible image interpretation [Pastukhov, Vonau, and Braun, 2012, Journal of Vision, 12(1), 1-16].

How subliminal visual input modulates visual short-term memory and imagery

Silvia Bona, Zaira Cattaneo, Tomaso Vecchi and Juha Silvanto

The interaction between visual short-term memory (VSTM) and visual perception has recently been the subject of much interest. Here we investigated how incoming perceptual information affects information held in VSTM. Subjects were asked to hold in memory the orientation of a grating; during the delay period, a visual distracter could appeared. The orientation of the distracter ranged from 0 to 40 degrees relative to the orientation of the memory stimulus. In the 'subliminal' condition, the distracter was followed by a mask, rendering it invisible; in the 'visible' condition, no mask was presented. In both cases, VSTM accuracy was impaired when the distracter orientation differed from the memory orientation by 10 or 40 degrees. No difference was found between the 'subliminal' and 'visible' conditions. In a second experiment, we investigated the influence of subliminal distracters on visual imagery by asking subjects to judge the subjective vividness of the memory stimulus at the end of each trial. Surprisingly, we found that the subliminal distracters reduced vividness not only when accuracy was impaired (with a memory-distracter difference of 40 deg), but also when the distracter was identical to the memory item and did not reduce accuracy.

Enforcing double dissociations between measures of priming and awareness (or anything else)

Thomas Schmidt

When comparing an indirect measure of visual perception with a direct measure of visual awareness, unconscious perception can be demonstrated by establishing an experimental manipulation that leads to opposite effects in the two measures, thus implying that they cannot both be explained by a single source of conscious information. Such a double dissociation, e.g. between priming and identification of a masked prime, requires none of the restrictive assumptions needed for demonstrating that a masked prime is 'invisible' [Schmidt & Vorberg, 2006, Perception & Psychophysics, 68, 489-504]. Here a psychophysical procedure is demonstrated that enforces double dissociations between masked priming and prime visibility by experimentally inducing qualitatively different time-courses of visual masking. Participants respond to a visual targets preceded by masked primes, and the intensity of the mask is systematically coupled to the prime-target SOA so that prime visibility either increases or decreases with SOA. A double dissociation is produced when motor priming effects keep increasing with SOA no matter whether prime visibility increases or decreases. Such 'enforced double dissociations' between priming and awareness are demonstrated for response priming and Eriksen paradigms. The technique can be applied to explore possible double dissociations in arbitrary fields under tight experimental control.

Interocular suppression eliminates the processing of perceptual ambiguity

Karin T. Ludwig, Veith A. Weilnhammer, Alexander Pastukhov, Philipp Sterzer and Guido Hesselmann

Ambiguous visual stimuli are often used as a unique window on perception and consciousness. It remains unclear, however, whether conscious perception is a necessary prerequisite for the processing of ambiguity. To address this question, we tested whether an ambiguous stimulus, rendered invisible by interocular suppression (continuous flash suppression, CFS), affects the perception of subsequently presented visible ambiguous stimuli. In a 2x2 experimental design, a Necker cube was presented either continuously or intermittently to one eye. CFS masks presented to the other eye were either present or absent. When present in the intermittent Necker cube condition, the CFS masks were shown during the stimulus-off periods. In the continuous condition, CFS masks were shown during the same time intervals as in the intermittent condition while the Necker cube was presented continuously. In this latter condition, the Necker cube was thus constantly present but intermittently suppressed from awareness by CFS. As could be shown by comparing distributions of dominance times, the number of perceptual switches, and the probability of perceptual reappearance of the last dominant percept across conditions, perception of the Necker cube remained unaffected by periods of invisible stimulus presentation. This suggests that perceptual ambiguity is not processed during interocular suppression.

Distinct Patterns of Spontaneous Fluctuations in Human Visual Cortex Reflect the Rate and Duration of Bistable Perceptual States

Tobias Donner, Dov Sagi, Yoram Bonneh and David Heeger

Models of bistable perceptual phenomena posit spontaneous fluctuations in cortical activity across multiple stages of the visual cortical hierarchy. We used fMRI to link activity fluctuations throughout multiple visual cortical areas to 'motion induced blindness' (MIB). MIB is a bistable phenomenon, in which a small salient target spontaneously disappears and reappears when surrounded by a moving mask. fMRI activity during MIB was measured in multiple retinotopic subregions corresponding to target and mask, within V1-V4. We computed a matrix of correlations between the fMRI time series in these sub-regions. We used singular value decomposition to identify spatial patterns of temporal correlations that reflected the dynamics of MIB. This identified all activity fluctuations, irrespective of whether or not these culminated in behavioral reports of MIB target disappearance. The fluctuations associated with MIB disappearance rate were target-specific in V4, but not V1. The fluctuations associated with the duration of MIB states were target-specific in V1, but not V4. These two distinct fluctuation patterns reflect distinct mechanisms contributing to MIB. We propose that the V4 fluctuations reflect competitive interactions that are closely coupled to behavioral report. Conversely, the V1 fluctuations may reflect adaptation dynamics that lay the foundation for target disappearance, decoupled from report.

Developing continuous flash suppression for moving stimuli

Pieter Moors, Johan Wagemans and Lee De-Wit

Continuous flash suppression (CFS) is a technique to reliably suppresses stimuli from visual awareness for extended periods of time [Tsuchiya and Koch, 2005, Nature Neuroscience, 8(8), 1096-1101]. Recently, researchers have begun to explore the properties that make a CFS mask effective. Yang and Blake [2012, Journal of Vision, 12(3):8, 1-14] report the importance of the spatial frequency match between the CFS mask and the suppressed stimulus. In this experiment, we present a mask that can effectively render moving stimuli unconscious. We tested the influence of the speed of mask elements on suppression strength. Our data show that the threshold for detecting a moving stimulus is highest when the speed of the individual elements of the mask matches the speed of the suppressed stimulus. The saliency maps for each mask show that there are larger bottom up transients in the classical CFS and faster masks while these prove to be less effective. This highlights that bottom up transients in salience are not as effective as matching the motion properties of the to-be-masked stimuli. These results stress the importance of constructing a CFS mask that matches certain features of the to be suppressed stimulus in order to reliably suppress it from awareness.

Abrupt color change of a previewed mask reinstates object substitution masking

Nobuyuki Hirose, Shouta Hattori and Shuji Mori

When a briefly presented target is surrounded by a sparse mask whose offset is delayed relative to the target offset, its identification is impaired compared to when both have common offset. This is a type of backward masking, called object substitution masking (OSM). Recent studies have reported mask preview effect, in which OSM is largely attenuated by mask preview. We investigated whether and how abrupt change in the color of a previewed mask affects mask preview effect. The basic effect that mask preview attenuates OSM was replicated. Abruptly changing the mask color at the target onset disrupted mask preview effect, that is, OSM was reinstated. In terms of the role of surface features in establishment and maintenance of object representations, we propose that a large color change allows the previewed mask to recompete with the target.

Top-down mechanisms and conscious visual experience

Alba Grieco and Armando Oliveira

In a previous study1 the effects of practice on a detection-task with contrast-polarity-cues were analyzed. The results showed evidence for distinct mechanisms underlying each target-condition (present-versus-absent), suggesting the general operation of center-surround-mechanisms with an additional involvement of top-down influences in the target-present-condition only. The role played by these feedback signals appears to be that of further inhibiting background elements and thereby enhancing the figure-ground processing. Here, the aim is to understand whether such top-down mechanisms are related to conscious perception. Ten participants rated their confidence regarding the presence-absence of the target in a four-point-scale. Using a perceptual-learning paradigm in combination with backward-masking, the results show that masking prevents improvement in both the objective (accuracy) and subjective (confidence rate) measures in the target-present condition, but leaves unchanged the learning effects in the target-absent condition, suggesting the involvement of feedback-loops in the enhancement of both detection accuracy and conscious awareness of the target. A dissociation found between the objective and subjective measures at the beginning of learning in the target-absent condition demonstrates that vision may guide behavior in the absence of conscious experience, as reported by several studies2,3, further suggesting that feed-forward mechanisms alone are not sufficient to promote conscious experience.

A direct model of indirect perception: the spread mind model

Riccardo Manzotti

One of the main arguments against direct perception and externalism by and large is the alleged evidence in favour of not veridical perception. In fact, it commonly held that there are many cases of indirect perception in which the external object is missing as it happens in dreams or illusions. The subject perceives something which seems to be not real. If this belief true, it would indeed be possible to have phenomenal visual experience without any corresponding external phenomenon. Contrary to such widespread view, here I will sketch a physical model of direct perception, dubbed the spread mind, that - by means of simple examples offered by optics such as mirror, transparent glasses, caleydoscopes - may endorse all alleged cases of indirect or not veridical perception (Manzotti 2006; Manzotti 2011a; Manzotti 2011b). I will review the relevant literature on indirect perception (afterimages, phosphenes in the congenitally blind, synaesthesia, migraine with aura, phantom limb) to show that alternative interpretation of data are available (Kirshfeld 1999; Brugger, Kollias et al. 2000; Cowey and Walsh 2000; Mulleners, Chronicle et al. 2001; Lopes da Silva 2003; Nir and Tononi 2009). To recap, I will try to show that all cases of perception might be reinterpreted as cases of direct and veridical perception.