Week 2: Visual Perception and Deception Flashcards
Three-stage model to help explain our eye’s and brain’s complex visual information processing system
Stage 1: Parallel processing to extract low-level properties of the visual scene.
Once the signal from our neurons in the eye reaches the visual cortex, networks of billions of neurons work in parallel to extract low-level properties of the visual field, such as colour, texture, orientation and movement. This happens automatically and unconsciously. The results of this process are held temporarily in iconic memory (<1000 ms), which is just enough time for our conscious processes to divert attention if required and move Attach Imagesto the next stage. During stage 1, information that is optimised for recognition by the vast neuronal networks in the visual cortex will be readily detected and processed. This will enable efficient interpretation of our data visualisation.
Stage 2: Pattern perception During the pattern perception stage, information obtained from stage 1 is further processed in order to identify patterns. Our visual field is broken into regions and our brains analyse contours, regions of colour, texture and motion. Stage 2 starts to involve our attention as we choose to make visual queries of the display. This stage is slower and less automatic than stage 1 but is still very rapid. Patterns perceived in the visual display can be held for a few seconds in our memory. Our brains will also transition between our object perception pathways and our action pathways. Data visualisation mainly concerns the object recognition pathway or the ‘what’ system. The action pathway relates to our body’s reaction to the environment. For example, the action of catching a ball rapidly approaching you.
Stage 3: Visual working memory
This is the fully conscious stage of visual information processing. Our attention has been drawn to a visual task, for example, interpreting a data visualisation that has caught our attention in an online news article. Our identification of the data visualisation based on the broad layout of the graphic is referred to as ‘gist’. We are all familiar with visual gist. Think about how quickly we can identify the scene of a TV show as you rapidly flick through channels. Almost instinctively, we can identify the broad spatial layout of the scene as something like a beach, park, bush, studio, house etc. This suggests another important implication for data visualisation. Using visualisations that have ‘gist’ will lead to more rapid interpretation. For example, most people are familiar with common data visualisations such as bar charts and scatter plots. Using these familiar data visualisation methods will allow the viewer to get the ‘gist’ of the data visualisation very quickly. Once we get the gist of the visualisation, we commence a series of visual queries driven by stages 1 and 2. We employ visual search strategies to bring the information together. Our working memory allows us to keep a few pieces of the visualisation in mind at any one time, however, we can also exploit our long-term memory stores to help fill in the gaps and activate contextual cues. With this in mind, Ware (2013) compiled the following list of the costs versus benefits considerations for data visualisation (p. 24–25).
Eight common Gestalt laws
Proximity
Objects close or clustering together are perceptually grouped. Data visualisations use proximity to highlight relationships between categories and trends in the data. This also means that spacing can be used to visualise no relationship. Network data visualisations such as “Transportation Clusters” by Grandjean (2016) use proximity to show clusters of nodes that share a relationship. The visualisation shows the increased connectedness of airports within continents which form distinct clusters, but also how airports connect between continents.
Similarity
Objects of similar characteristics (e.g. size, shape, colour) are grouped. Visualisation implication uses colour, size, shape and other attributes to group related objects or to differentiate between categories.
Connectedness
Connectedness is more powerful than proximity, colour, size or shape. Objects connected by lines demonstrate relationships between objects. Data visualisations, such as times series plots and network diagrams, use connections between data points to represent temporal changes and highlight relationships.
Continuity
This law predicts that we are inclined to perceive objects from elements that are smooth and continuous, versus irregular and jagged. Smooth lines are easier to perceive the connection between data points and identify trends. We should also be careful with how we arrange foreground and background objects that overlap to ensure objects are perceived correctly.
Symmetry
We tend to group symmetrical objects together. Aligning data visualisations, for example side-by-side, promotes comparisons and allows differences to be readily perceived.
Closure
Closure refers to our tendency to ‘fill in the gaps’ when we see incomplete patterns that resemble familiar shapes and objects. This means we should group related objects around closed shapes and ensure that overlapping objects are ‘closed’ correctly by the brain. Contrasting overlapping objects using shapes and colour can help ensure the correct closure is achieved.
Figure-ground principle
The figure-ground effect tells us that smaller objects within a figure are interpreted as the foreground, while larger objects make up the background. Data visualisations often plot data objects to backgrounds. This ensures that the objects within the border of the background are perceived to be representations of the data.
Common fate
Objects perceived to be moving in the same direction are grouped together and share a common path. Animated and interactive data visualisations use this principle to show relationships between categories and highlight changes across time or based on user input.
