Periodic Reporting for period 4 - ContentMAP (Contentotopic mapping: the topographical organization of object knowledge in the brain)
Reporting period: 2023-08-01 to 2025-03-31
Our ability to recognize an object amongst many others is one of the most important features of the human mind. However, object recognition requires tremendous computational effort, as we need to solve a complex and recursive environment with ease and proficiency. This challenging feat is dependent on the implementation of an effective organization of knowledge in the brain. In ContentMAP I will put forth a novel understanding of how object knowledge is organized in the brain, by proposing that this knowledge is topographically laid out in the cortical surface according to object-related dimensions that code for different types of representational content – I will call this contentotopic mapping. To study this fine-grain topography, I will use a combination of fMRI, behavioral, and neuromodulation approaches. I will first obtain patterns of neural and cognitive similarity between objects, and from these extract object-related dimensions using a dimensionality reduction technique. I will then parametrically manipulate these dimensions with an innovative use of a visual field mapping technique, and test how functional selectivity changes across the cortical surface according to an object’s score on a target dimension. Moreover, I will test the tuning function of these contentotopic maps. Finally, to mirror the complexity of implementing a highdimensional manifold onto a 2D cortical sheet, I will aggregate the topographies for the different dimensions into a composite map, and develop an encoding model to predict neural signatures for each object. To sum up, ContentMAP will have a dramatic impact in the cognitive sciences by describing how the stuff of concepts is represented in the brain, and providing a complete description of how fine-grain representations and functional selectivity within highlevel complex processes are topographically implemented.
ContentMAP tries to address how information about (manipulable) objects is organized in the brain. Specifically, ContentMAP proposes that, in part, the organization of object knowledge followed the typical organizational principles that the brain applies elsewhere – i.e. that object information is topographically organized by (object-related) dimensions. During this first part of ContentMAP, we have used the object-related dimensions we had obtained (and described in the Action), and tested whether these dimensions: a) are organizing principles for neural data (as they are for behavioral judgements); and b) whether that organization is topographic – i.e. are objects represented in different areas in neural proximity of each other based on whether they are similar in these object-related dimensions?
Our preliminary data suggests that 1) our object-related dimensions are used as organizing principles for neural data, in that decoding of object-specific neural patterns is influenced by object-specific score in these dimensions; and that 2) these dimensions drive a topography organization of information - what we call contentotopy, in that when we use visual mapping techniques such as population receptive field we obtain continuous maps in different areas. We have also developed a parallel (and not originally proposed in the Action) line of research focusing also on the organization of object knowledge in the brain and particularly on the role of connectivity in how conceptual information is processed and organized. Here, we have shown that object-related local computations are shaped by long-range connectivity with regions that share high-level object preferences in order to fulfill particular cognitive demands.
Our preliminary data suggests that 1) our object-related dimensions are used as organizing principles for neural data, in that decoding of object-specific neural patterns is influenced by object-specific score in these dimensions; and that 2) these dimensions drive a topography organization of information - what we call contentotopy, in that when we use visual mapping techniques such as population receptive field we obtain continuous maps in different areas. We have also developed a parallel (and not originally proposed in the Action) line of research focusing also on the organization of object knowledge in the brain and particularly on the role of connectivity in how conceptual information is processed and organized. Here, we have shown that object-related local computations are shaped by long-range connectivity with regions that share high-level object preferences in order to fulfill particular cognitive demands.
We proposed to unravel how fine-grain representational content is represented in the brain by a) uncovering object-related dimensions that drive the fine-grain topography of object-related areas; and b) testing how these dimensions shape our interactions with objects. We first focused on finding object-related dimensions, and exploring the representational content of object knowledge. We show that object representation follows multidimensionality. Almeida et al. (2023; CommsBio) obtained a series of dimensions, extracted from how individuals mentally structure manipulable object knowledge through a dimensionality reduction technique. Specifically, we show that dimensions such as object material, object elongation, or grasp type, among others, structure our mental reasoning about (manipulable) objects. Importantly, here we also show that these dimensions are able to predict neural response to the presentations of the objects themselves. We further explore this aspect of object related dimensionality in several other papers. For instance, in Bergstrom et al. (2021; Cortex) and Hussain et al., (2024; Neuropsychologia) we demonstrate the importance of the dimension “grasp type” in how information about objects is stored and represented in the brain. We also showed that the extracted dimensions can predict behavioral responses (Almeida et al., 2023; CommsBio), even for items that were not used to extract those dimensions (Walbrin et al., 2024; iSCIENCE).
We then showed that the different object-related dimensions obtained previously governed the topography of object knowledge in the brain. Importantly, and for the first time, we showed topographical maps for object-related dimensions in dorsal and ventral occipital cortex that code for the score of each object on each target dimension in a linear progression following a particular direction along the cortical surface. Maps for each dimension are distinct, are consistent across individuals, and are not exhausted by eccentricity (i.e. major low-level visual confounds). Thus, object information is coded in multiple topographical maps – i.e. contentopic maps. These contentopic maps refer to intermediate level visual and visuomotor representations.
We then showed that the different object-related dimensions obtained previously governed the topography of object knowledge in the brain. Importantly, and for the first time, we showed topographical maps for object-related dimensions in dorsal and ventral occipital cortex that code for the score of each object on each target dimension in a linear progression following a particular direction along the cortical surface. Maps for each dimension are distinct, are consistent across individuals, and are not exhausted by eccentricity (i.e. major low-level visual confounds). Thus, object information is coded in multiple topographical maps – i.e. contentopic maps. These contentopic maps refer to intermediate level visual and visuomotor representations.