Feature Distribution Learning (FDL): A New Method for Studying Visual Ensembles Perception with Priming of Attention Shifts

Memory reports are biased by all relevant contents of working memory

Sensory input is inherently noisy while the world is inherently predictable. When multiple observations of the same object are available, integration of the available information necessarily increases the reliability of a world estimate. Optimal integration of multiple instances of sensory evidence has already been demonstrated during multisensory perception but could benefit unimodal perception as well. In the present study 330 participants observed a sequence of four orientations and were cued to report one of them. Reports were biased by all simultaneously memorized items that were similar and relevant to the target item, weighted by their reliability (signal-to-noise ratio). Orientations presented before and presented after the target biased report, demonstrating that the bias emerges in memory and not (exclusively) during perception or encoding. Only attended, task-relevant items biased report. We suggest that these results reflect how the visual system integrates information that is sampled from the same object at consecutive timepoints to promote perceptual stability and behavioural effectiveness in a dynamic world. We suggest that similar response biases, such as serial dependence, might be instances of a more general mechanism of working memory averaging. Data is available at https://osf.io/embcf/ .

The influence of the tested item on serial dependence in perceptual decisions

Serial dependence in vision reflects how perceptual decisions can be biased by what we have recently perceived. Serial dependence studies test single items' effects on perceptual decisions. However, our visual world contains multiple objects at any given moment, so it's important to understand how past experiences affect not only a single object but also perception in a more general sense. Here we asked the question: What effect does a single item have when there is more than one subsequently presented test item? We displayed a single line (inducer) at the screen center, then either a single test-line or two simultaneous test-lines, varying in orientation space to the inducer. Next, participants reported test-line orientation using a left or right located response circle (to indicate which test-line should be reported). The results demonstrated that the inducer influenced subsequent perceptual judgments of a test-line even when two test-lines were presented. Distant items caused repulsive serial dependence, while close items caused attractive serial dependence. This shows how a single inducer can influence test-line judgments, even when two test-lines are presented, and can produce attractive and repulsive serial dependence biases when the item to report is revealed after it has disappeared.

Priming of probabilistic attentional templates

Attentional priming has a dominating influence on vision, speeding visual search, releasing items from crowding, reducing masking effects, and during free-choice, primed targets are chosen over unprimed ones. Many accounts postulate that templates stored in working memory control what we attend to and mediate the priming. But what is the nature of these templates (or representations)? Analyses of real-world visual scenes suggest that tuning templates to exact color or luminance values would be impractical since those can vary greatly because of changes in environmental circumstances and perceptual interpretation. Tuning templates to a range of the most probable values would be more efficient. Recent evidence does indeed suggest that the visual system represents such probability, gradually encoding statistical variation in the environment through repeated exposure to input statistics. This is consistent with evidence from neurophysiology and theoretical neuroscience as well as computational evidence of probabilistic representations in visual perception. I argue that such probabilistic representations are the unit of attentional priming and that priming of, say, a repeated single-color value simply involves priming of a distribution with no variance. This "priming of probability" view can be modelled within a Bayesian framework where priming provides contextual priors. Priming can therefore be thought of as learning of the underlying probability density function of the target or distractor sets in a given continuous task.

Ensemble averaging: What can we learn from skewed feature distributions?

Many studies have shown that observers can accurately estimate the average feature of a group of objects. However, the way the visual system relies on the information from each individual item is still under debate. Some models suggest some or all items sampled and averaged arithmetically. Another strategy implies "robust averaging," when middle elements gain greater weight than outliers. One version of a robust averaging model was recently suggested by Teng et al. (2021), who studied motion direction averaging in skewed feature distributions and found systematic biases toward their modes. They interpreted these biases as evidence for robust averaging and suggested a probabilistic weighting model based on minimization of the virtual loss function. In four experiments, we replicated systematic skew-related biases in another feature domain, namely, orientation averaging. Importantly, we show that the magnitude of the bias is not determined by the locations of the mean or mode alone, but is substantially defined by the shape of the whole feature distribution. We test a model that accounts for such distribution-dependent biases and robust averaging in a biologically plausible way. The model is based on well-established mechanisms of spatial pooling and population encoding of local features by neurons with large receptive fields. Both the loss functions model and the population coding model with a winner-take-all decoding rule accurately predicted the observed patterns, suggesting that the pooled population response model can be considered a neural implementation of the computational algorithms of information sampling and robust averaging in ensemble perception.

Unlocking crowding by ensemble statistics

In crowding,^{1,2,3,4,5,6,7} objects that can be easily recognized in isolation appear jumbled when surrounded by other elements.⁸ Traditionally, crowding is explained by local pooling mechanisms,^{3,6,9,10,11,12,13,14,15} but many findings have shown that the global configuration of the entire stimulus display, rather than local aspects, determines crowding.^{8,16,17,18,19,20,21,22,23,24,25,26,27,28} However, understanding global configurations is challenging because even slight changes can lead from crowding to uncrowding and vice versa.^23,25,28,29 Unfortunately, the number of configurations to explore is virtually infinite. Here, we show that one does not need to know the specific configuration of flankers to determine crowding strength but only their ensemble statistics, which allow for the rapid computation of groups within the stimulus display.^{30,31,32,33,34,35,36,37} To investigate the role of ensemble statistics in (un)crowding, we used a classic vernier offset discrimination task in which the vernier was flanked by multiple squares. We manipulated the orientation statistics of the squares based on the following rationale: a central square with an orientation different from the mean orientation of the other squares stands out from the rest and groups with the vernier, causing strong crowding. If, on the other hand, all squares group together, the vernier is the only element that stands out, and crowding is weak. These effects should depend exclusively on the perceived ensemble statistics, i.e., on the mean orientation of the squares and not on their individual orientations. In two experiments, we confirmed these predictions.

What kind of empirical evidence is needed for probabilistic mental representations? An example from visual perception

Recent accounts of perception and cognition propose that the brain represents information probabilistically. While this assumption is common, empirical support for such probabilistic representations in perception has recently been criticized. Here, we evaluate these criticisms and present an account based on a recently developed psychophysical methodology, Feature Distribution Learning (FDL), which provides promising evidence for probabilistic representations by avoiding these criticisms. The method uses priming and role-reversal effects in visual search. Observers' search times reveal the structure of perceptual representations, in which the probability distribution of distractor features is encoded. We explain how FDL results provide evidence for a stronger notion of representation that relies on structural correspondence between stimulus uncertainty and perceptual representations, rather than a mere co-variation between the two. Moreover, such an account allows us to demonstrate what kind of empirical evidence is needed to support probabilistic representations as posited in current probabilistic Bayesian theories of perception.

Testing temporal integration of feature probability distributions using role-reversal effects in visual search

The visual system is sensitive to statistical properties of complex scenes and can encode feature probability distributions in detail. But does the brain use these statistics to build probabilistic models of the ever-changing visual input? To investigate this, we examined how observers temporally integrate two different orientation distributions from sequentially presented visual search trials. If the encoded probabilistic information is used in a Bayesian optimal way, observers should weigh more reliable information more strongly, such as feature distributions with low variance. We therefore manipulated the variance of the two feature distributions. Participants performed sequential odd-one-out visual search for an oddly oriented line among distractors. During successive learning trials, the distractor orientations were sampled from two different Gaussian distributions on alternating trials. Then, observers performed a 'test trial' where the orientations of the target and distractors were switched, allowing us to assess observer's internal representation of distractor distributions based on changes in response times. In three experiments we observed that observer's search times on test trials depended mainly on the very last learning trial, indicating a strong recency effect. Since temporal integration has been previously observed with this method, we conclude that when the input is unreliable, the visual system relies more on the most recent stimulus. This indicates that the visual system prefers to utilize sensory history when the statistical properties of the environment are relatively stable.

Target templates in low target-distractor discriminability visual search have higher resolution, but the advantage they provide is short-lived

When you search repeatedly for a set of items among very similar distractors, does that make you more efficient in locating the targets? To address this, we had observers search for two categories of targets among the same set of distractors across trials. Visual and conceptual similarity of the stimuli were validated with a multidimensional scaling analysis, and separately using a deep neural network model. After a few blocks of visual search trials, the distractor set was replaced. In three experiments, we manipulated the level of discriminability between the targets and distractors before and after the distractors were replaced. Our results suggest that in the presence of repeated distractors, observers generally become more efficient. However, the difficulty of the search task does impact how efficient people are when the distractor set is replaced. Specifically, when the training is easy, people are more impaired in a difficult transfer test. We attribute this effect to the precision of the target template generated during training. In particular, a coarse target template is created when the target and distractors are easy to discriminate. These coarse target templates do not transfer well in a context with new distractors. This suggests that learning with more distinct targets and distractors can result in lower performance when context changes, but observers recover from this effect quickly (within a block of search trials).

Keeping it real: Looking beyond capacity limits in visual cognition

Research within visual cognition has made tremendous strides in uncovering the basic operating characteristics of the visual system by reducing the complexity of natural vision to artificial but well-controlled experimental tasks and stimuli. This reductionist approach has for example been used to assess the basic limitations of visual attention, visual working memory (VWM) capacity, and the fidelity of visual long-term memory (VLTM). The assessment of these limits is usually made in a pure sense, irrespective of goals, actions, and priors. While it is important to map out the bottlenecks our visual system faces, we focus here on selected examples of how such limitations can be overcome. Recent findings suggest that during more natural tasks, capacity may be higher than reductionist research suggests and that separable systems subserve different actions, such as reaching and looking, which might provide important insights about how pure attentional or memory limitations could be circumvented. We also review evidence suggesting that the closer we get to naturalistic behavior, the more we encounter implicit learning mechanisms that operate "for free" and "on the fly." These mechanisms provide a surprisingly rich visual experience, which can support capacity-limited systems. We speculate whether natural tasks may yield different estimates of the limitations of VWM, VLTM, and attention, and propose that capacity measurements should also pass the real-world test within naturalistic frameworks. Our review highlights various approaches for this and suggests that our understanding of visual cognition will benefit from incorporating the complexities of real-world cognition in experimental approaches.

Optimizing perception: Attended and ignored stimuli create opposing perceptual biases

Humans have remarkable abilities to construct a stable visual world from continuously changing input. There is increasing evidence that momentary visual input blends with previous input to preserve perceptual continuity. Most studies have shown that such influences can be traced to characteristics of the attended object at a given moment. Little is known about the role of ignored stimuli in creating this continuity. This is important since while some input is selected for processing, other input must be actively ignored for efficient selection of the task-relevant stimuli. We asked whether attended targets and actively ignored distractor stimuli in an odd-one-out search task would bias observers' perception differently. Our observers searched for an oddly oriented line among distractors and were occasionally asked to report the orientation of the last visual search target they saw in an adjustment task. Our results show that at least two opposite biases from past stimuli influence current perception: A positive bias caused by serial dependence pulls perception of the target toward the previous target features, while a negative bias induced by the to-be-ignored distractor features pushes perception of the target away from the distractor distribution. Our results suggest that to-be-ignored items produce a perceptual bias that acts in parallel with other biases induced by attended items to optimize perception. Our results are the first to demonstrate how actively ignored information facilitates continuity in visual perception.