Reading as a visual act: Recognition of visual letter symbols in the mind and brain

Time/Room: Friday, May 17, 2019, 12:00 – 2:00 pm, Talk Room 1
Organizer(s): Teresa Schubert, Harvard University
Presenters: Teresa Schubert, Alex Holcombe, Kalanit Grill-Spector, Karin James
< Back to 2019 Symposia

Symposium Description

A large proportion of our time as literate adults is spent reading: Deriving meaning from visual symbols. Letter symbols have only been in use for a few millennia; our visual system, which may have evolved to recognize lions and the faces of our kin, is now required to recognize the written word “LION” and the handwriting of your nephew. How does the visual system accomplish this unique feat of recognition? A wealth of studies consider early visual abilities that are involved in letter recognition but the study of these symbols as visual objects is relatively rare. In this symposium, we will highlight work by a growing number of researchers attempting to bridge the gap in research between vision and language by investigating letter and word recognition processes. In addition to interest in reading on its own merits, we propose that a minimal understanding of letter recognition is relevant to vision scientists in related domains. Many popular paradigms, from visual search to the attentional blink, use letters as stimuli. Letters are also a unique class within visual objects, and an understanding of these stimuli can constrain broader theories. Furthermore, letters can be used as a comparison class to other stimuli with which humans have high levels of expertise, such as faces and tools. In this symposium, we will discuss the state of the science of letter recognition from both a cognitive and neural perspective. We will provide attendees with information specific to letter/word recognition and situate these findings relative to broader visual cognition. Our speakers span the range from junior to established scientists and use both behavioral and neural approaches. In the first talk, Schubert will present an overview of letter recognition, describing the hierarchical stages of abstraction and relating them to similar stages proposed in object recognition. In the second talk, Holcombe will address the relationship between domain-general abilities and letter recognition, by manipulating orthographic properties such as reading direction to interrogate capacity limits and laterality effects in visual working memory. In the third talk, Grill-Spector will discuss how foveal visual experience with words contributes to the organization of ventral temporal cortex over development. In the fourth talk, James will discuss the relationship between letter recognition and letter production. In addition to their visual properties letters have associated motor plans for production, and she will present evidence suggesting this production information may be strongly linked to letter recognition. Finally, we will integrate these levels into a discussion of broad open questions in letter recognition that have relevance across visual perception, such as: What are the limits of the flexibility of visual recognition systems? At what level do capacity limits in memory encoding operate? What pressures give rise to the functional organization of ventral temporal cortex? What is the extent of interactions between systems for visual perception and for motor action? On the whole, we anticipate that this symposium will provide a new perspective on the study of letter recognition and its relevance to work across the range of visual cognition.


How do we recognize letters as visual objects?

Speaker: Teresa Schubert, Harvard University
Additional Authors: David Rothlein, VA Boston Healthcare System; Brenda Rapp, Johns Hopkins University

How do we recognize b and B as instances of the same letter? The cognitive mechanisms of letter recognition permit abstraction across highly different visual exemplars of the same letter (b and B), while also differentiating between highly similar exemplars of different letters (c and e). In this talk, I will present a hierarchical framework for letter recognition which involves progressively smaller reliance on sensory stimulus details to achieve abstract letter representation. In addition to abstraction across visual features, letter recognition in this framework also involves different levels of abstraction in spatial reference frames. This theory was developed based on data from individuals with acquired letter identification deficits (subsequent to brain lesion) and further supported by behavioral and neural research with unimpaired adult readers. I will relate this letter recognition theory to the seminal Marr & Nishihara (1978) framework for object recognition, arguing that letter recognition and visual object recognition require a number of comparable computations, leading to broadly similar recognition systems. Finally, I will compare and contrast neural evidence of cross-modal (visual and auditory letter name) representations for letters and objects. Overall, this talk will provide a theoretical and empirical framework within which to consider letter recognition as a form of object recognition.

Implicit reading direction and limited-capacity letter identification

Speaker: Alex Holcombe, University of Sydney
Additional Authors: Kim Ransley, University of Sydney

Reading this sentence was quite an accomplishment. You overcame a poor ability, possibly even a complete inability, to simultaneously identify multiple objects – according to the influential “EZ reader” model of reading, humans can identify only one word at a time. In the field of visual attention, it is known that if one must identify multiple simultaneously-presented stimuli, spatial biases may be present but are often small. Reading a sentence, by contrast, involves a highly stereotyped attentional routine with rapid but serial, or nearly serial, identification of stimuli from left to right. Unexpectedly, my lab has found evidence that this reading routine is elicited when just two widely-spaced letters are briefly presented and observers are asked to identify both letters. We find a large left-side performance advantage that is absent or reversed when the two letters are rotated to face to the left instead of to the right. Additional findings from RSVP (rapid serial visual presentation) lead us to suggest that both letters are selected by attention simultaneously, with the bottleneck at which one letter is prioritized sitting at a late stage of processing – identification or working memory consolidation. Thus, a rather minimal cue of letter orientation elicits a strong reading direction-based prioritization routine, which will allow better understanding of both the bottleneck in visual identification and how reading overcomes it.

How learning to read affects the function and structure of ventral temporal cortex

Speaker: Kalanit Grill-Spector, Stanford University
Additional Authors: Marisa Nordt, Stanford University; Vaidehi Natu, Stanford University; Jesse Gomez, Stanford University and UC Berkeley; Brianna Jeska, Stanford University; Michael Barnett, Stanford University

Becoming a proficient reader requires substantial learning over many years. However, it is unknown how learning to read affects development of distributed visual representations across human ventral temporal cortex (VTC). Using fMRI and a data-driven approach, we examined if and how distributed VTC responses to characters (pseudowords and numbers) develop after age 5. Results reveal anatomical- and hemisphere-specific development. With development, distributed responses to words and characters became more distinctive and informative in lateral but not medial VTC, in the left, but not right, hemisphere. While development of voxels with both positive and negative preference to characters affected distributed information, only activity across voxels with positive preference to characters correlated with reading ability. We also tested what developmental changes occur to the gray and white matter, by obtaining in the same participants quantitative MRI and diffusion MRI data. T1 relaxation time from qMRI and mean diffusivity (MD) from dMRI provide independent measurements of microstructural properties. In character-selective regions in lateral VTC, but not in place-selective regions in medial VTC, we found that T1 and MD decreased from age 5 to adulthood, as well as in their adjacent white matter. T1 and MD decreases are consistent with tissue growth and were correlated with the apparent thinning of lateral VTC. These findings suggest the intriguing possibility that regions that show a protracted functional development also have a protracted structural development. Our data have important ramifications for understanding how learning to read affects brain development, and for elucidating neural mechanisms of reading disabilities.

Visual experiences during letter production contribute to the development of the neural systems supporting letter perception

Speaker: Karin James, Indiana University
Additional Authors: Sophia Vinci-Booher, Indiana University

Letter production is a perceptual-motor activity that creates visual experiences with the practiced letters. Past research has focused on the importance of the motor production component of writing by hand, with less emphasis placed on the potential importance of the visual percepts that are created. We sought to better understand how different visual percepts that result from letter production are processed at different levels of literacy experience. During fMRI, three groups of participants, younger children, older children, and adults, ranging in age from 4.5 to 22 years old, were presented with dynamic and static re-presentations of their own handwritten letters, static presentations of an age-matched control’s handwritten letters, and typeface letters. In younger children, we found that only the ventral-temporal cortex was recruited, and only for handwritten forms. The response in the older children also included only the ventral-temporal cortex but was associated with both handwritten and typed letter forms. The response in the adults was more distributed than in the children and responded to all types of letter forms. Thus, the youngest children processed exemplars, but not letter categories in the VTC, while older children and adults generalized their processing to many letter forms. Our results demonstrate the differences in the neural systems that support letter perception at different levels of experience and suggest that the perception of handwritten forms is an important component of how letter production contributes to developmental changes in brain processing

<Back to 2019 Symposia