Reverse-engineering the brain involves adopting and testing a hierarchy of working hypotheses regarding the computational problems that it solves, the representations and algorithms that it employs and the manner in which these are implemented. Because problem-level assumptions set the course for the entire research programme, it is particularly important to be open to the possibility that we have them wrong, but tacit algorithm- and implementation-level hypotheses can also benefit from occasional scrutiny. This paper focuses on the extent to which our computational understanding of how the brain works is shaped by three such rarely discussed assumptions, which span the levels of Marr's hierarchy: (i) that animal behaviour amounts to a series of stimulus/response bouts, (ii) that learning can be adequately modelled as being driven by the optimisation of a fixed objective function and (iii) that massively parallel, uniformly connected layered or recurrent network architectures suffice to support learning and behaviour. In comparison, a more realistic approach acknowledges that animal behaviour in the wild is characterised by dynamically branching serial order and is often agentic rather than reactive. Arguably, such behaviour calls for open-ended learning of world structure and may require a neural architecture that includes precisely wired circuits reflecting the serial and branching structure of behavioural tasks.

Many works in computer vision attempt to solve different tasks such as object detection, scene recognition or attribute detection, either separately or as a joint problem. In recent years, there has been a growing interest in combining the results from these different tasks in order to provide a textual description of the scene. However, when describing a scene, there are many items that can be mentioned. If we include all the objects, relationships, and attributes that exist in the image, the description would be extremely long and not convey a true understanding of the image. We present a novel approach to ranking the importance of the items to be described. Specifically, we focus on the task of discriminating one image from a group of others. We investigate the factors that contribute to the most efficient description that achieves this task. We also provide a quantitative method to measure the description quality for this specific task using data from human subjects and show that our method achieves better results than baseline methods.

Are minds really dynamical or are they really symbolic? Because minds are bundles of computations, and because computation is always a matter of interpretation of one system by another, minds are necessarily symbolic. Because minds, along with everything else in the universe, are physical, and insofar as the laws of physics are dynamical, minds are necessarily dynamical systems. Thus, the short answer to the opening question is ‘yes’. It makes sense to ask further whether some of the computations that constitute a human mind are constrained by functional, algorithmic, or implementational factors to be essentially of the discrete symbolic variety (even if they supervene on an apparently continuous dynamical substrate). I suggest that here too the answer is ‘yes’ and discuss the need for such discrete, symbolic cognitive computations in communication-related tasks.

The three commentaries of Van Orden, Spivey and Anderson, and Dietrich (with Markman's as a backdrop) form a tableau that reminds me of a fable by Ivan Andreevich Krylov (1769–1844), in which a swa...

January 01 2006 Book Review Mostly Harmless: Action in Perception. AlvaNoë. ( 2005, MIT Press.) $38.00 (Cloth), 392 pp. Shimon Edelman Shimon Edelman Department of Psychology, 232 Uris Hall, Cornell University, Ithaca, NY 14853-7601 [email protected] Search for other works by this author on: This Site Google Scholar Author and Article Information Shimon Edelman Department of Psychology, 232 Uris Hall, Cornell University, Ithaca, NY 14853-7601 [email protected] Online Issn: 1530-9185 Print Issn: 1064-5462 © 2006 Massachusetts Institute of Technology2006 Artificial Life (2006) 12 (1): 183–186. https://doi.org/10.1162/106454606775186482 Cite Icon Cite Permissions Share Icon Share Twitter LinkedIn Views Icon Views Article contents Figures & tables Video Audio Supplementary Data Peer Review Search Site Citation Shimon Edelman; Book Review. Artif Life 2006; 12 (1): 183–186. doi: https://doi.org/10.1162/106454606775186482 Download citation file: Ris (Zotero) Reference Manager EasyBib Bookends Mendeley Papers EndNote RefWorks BibTex toolbar search Search nav search search input Search input auto suggest search filter All ContentAll JournalsArtificial Life Search Advanced Search This content is only available as a PDF. © 2006 Massachusetts Institute of Technology2006 Article PDF first page preview Close Modal You do not currently have access to this content.

We describe a pattern acquisition algorithm that learns, in an unsuper- vised fashion, a streamlined representation of linguistic structures from a plain natural-language corpus. This paper addresses the issues of learn- ing structured knowledge from a large-scale natural language data set, and of generalization to unseen text. The implemented algorithm repre- sents sentences as paths on a graph whose vertices are words (or parts of words). Significant patterns, determined by recursive context-sensitive statistical inference, form new vertices. Linguistic constructions are rep- resented by trees composed of significant patterns and their associated equivalence classes. An input module allows the algorithm to be sub- jected to a standard test of English as a Second Language (ESL) profi- ciency. The results are encouraging: the model attains a level of per- formance considered to be “intermediate” for 9th-grade students, de- spite having been trained on a corpus (CHILDES) containing transcribed speech of parents directed to small children.

The distributional principle according to which morphemes that occur in identical contexts belong, in some sense, to the same category [1] has been advanced as a means for extracting syntactic structures from corpus data. We extend this principle by applying it recursively, and by us- ing mutual information for estimating category coherence. The resulting model learns, in an unsupervised fashion, highly structured, distributed representations of syntactic knowledge from corpora. It also exhibits promising behavior in tasks usually thought to require representations anchored in a grammar, such as systematicity.

To find out how the representations of structured visual objects depend on the co-occurrence statistics of their constituents, we exposed subjects to a set of composite images with tight control exerted over (1) the condi- tional probabilities of the constituent fragments, and (2) the value of Bar- low’s criterion of “suspicious coincidence” (the ratio of joint probability to the product of marginals). We then compared the part verification re- sponse times for various probe/target combinations before and after the exposure. For composite probes, the speedup was much larger for tar- gets that contained pairs of fragments perfectly predictive of each other, compared to those that did not. This effect was modulated by the sig- nificance of their co-occurrence as estimated by Barlow’s criterion. For lone-fragment probes, the speedup in all conditions was generally lower than for composites. These results shed light on the brain’s strategies for unsupervised acquisition of structural information in vision.

We describe a unified framework for the understanding of struc(cid:173) ture representation in primate vision. A model derived from this framework is shown to be effectively systematic in that it has the ability to interpret and associate together objects that are related through a rearrangement of common "middle-scale" parts, repre(cid:173) sented as image fragments. The model addresses the same concerns as previous work on compositional representation through the use of what+where receptive fields and attentional gain modulation. It does not require prior exposure to the individual parts, and avoids the need for abstract symbolic binding. 1 The problem of structure representation The focus of theoretical discussion in visual object processing has recently started to shift from problems of recognition and categorization to the representation of object structure. Although view- or appearance-based solutions for these problems proved effective on a variety of object classes [1], the "holistic" nature of this approach - the lack of explicit representation of relational structure - limits its appeal as a general framework for visual representation [2]. The main challenges in the processing of structure are productivity and system(cid:173) aticity, two traits commonly attributed to human cognition. A visual system is productive if it is open-ended, that is, if it can deal effectively with a potentially infinite set of objects. A visual representation is systematic if a well-defined change in the spatial configuration of the object (e.g., swapping top and bottom parts) causes a principled change in the representation (e.g., the interchange of the rep(cid:173) resentations of top and bottom parts [3, 2]). A solution commonly offered to the twin problems of productivity and systematicity is compositional representation, in which symbols standing for generic parts drawn from a small repertoire are bound together by categorical symbolically coded relations [4]. 2 The Chorus of Fragments In visual representation, the need for symbolic binding may be alleviated by us(cid:173) ing location in the visual field in lieu of the abstract frame that encodes object structure. Intuitively, the constituents of the object are then bound to each other by virtue of residing in their proper places in the visual field; this can be thought of as a pegboard, whose spatial structure supports the arrangement of parts sus(cid:173) pended from its pegs. This scheme exhibits shallow compositionality, which can be enhanced by allowing the "pegboard" mechanism to operate at different spatial scales, yielding effective systematicity across levels of resolution. Coarse coding the constituents (e.g., representing each object fragment in terms of its similarities to some basis shapes) will render the scheme productive. We call this approach to the representation of structure the Chorus of Fragments (CoF; [5]). 2.1 Neurobiological building blocks What+ Where cells. The representation of spatially anchored object fragments pos(cid:173) tulated by the CoF model can be supported by what+where neurons, each tuned both to a certain shape class and to a certain range of locations in the visual field. Such cells have been found in the monkey in areas V 4 and posterior IT [6], and in the prefrontal cortex [7]. Attentional gain fields. To decouple the representation of object structure from its location in the visual field, one needs a version of the what+where mechanism in which the response of the cell depends not merely on the location of the stimulus with respect to fixation (as in classical receptive fields), but also on its location with respect to the focus of attention. Indeed, modulatory effects of object-centered attention on classical RF structure (gain fields) have been found in area V 4 [8]. 2.2 Implemented model Our implementation of the CoF model involves what+where cells with attention(cid:173) modulated gain fields, and is aimed at productive and systematic treatment of composite shapes in object-centered coordinates. It operates directly on gray-level images, pre-processed by a model of the primary visual cortex [9], with complex(cid:173) cell responses modified to use the MAX operation suggested in [10]. In the model, one what+where unit is assigned to the top and one to the bottom fragment of the visual field, each extracted by an appropriately configured Gaussian gain profile (Figure 2, left). The units are trained (1) to discriminate among five objects, (2) to tolerate translation within the hemifield, and (3) to provide an estimate of the reliability of its output, through an autoassociation mechanism attempting to reconstruct the stimulus image [11, 12]. Within each hemifield, the five outputs of a unit can provide a coarse coding of novel objects belonging to the familiar category, in a manner useful for translation-tolerant recognition [13]. The reliability estimate carries information about category, allowing outputs for objects from other categories to be squelched. Most importantly, due to the spatial localization of the unit's receptive field, the system can distinguish between different configurations of the same shapes, while noting the fragment-wise similarities. We assume that during learning the system performs multiple fixations of the target object, effectively providing the what+where units with a basis for spanning the

We present a general scheme for learning sensorimotor tasks, which allows rapid online learning and generalization of the learned knowledge to unfamiliar objects. The scheme consists of two modules, the first generating candidate actions and the second estimating their quality. Both modules work in an alternating fashion until an action which is expected to provide satisfactory performance is generated, at which point the system executes the action. We developed a method for off-line selection of heuristic strategies and quality predicting features, based on statistical analysis. The usefulness of the scheme was demonstrated in the context of learning visually guided grasping. We consider a system that coordinates a parallel-jaw gripper and a fixed camera. The system learns to estimate grasp quality by learning a function from relevant visual features to the quality. An experimental setup using an AdeptOne manipulator was developed to test the scheme.

A representational scheme under which the ranking between represented similarities is isomorphic to the ranking between the corresponding shape similarities can support perfectly correct shape classification because it preserves the clustering of shapes according to the natural kinds prevailing in the external world. This article discusses the computational requirements of representation that preserves similarity ranks and points out the relative straightforwardness of its connectionist implementation.

We present an analysis of the effects of face distinctiveness on the performance of a computational model of recognition over viewpoint change. In the first stage of the model, the face stimulus is normalized by being mapped to an arbitrary standard view. In the second stage, the normalized stimulus is mapped into a "face space" spanned by a number of reference faces, and is classified as familiar or unfamiliar We carried out experiments employing a parametrically generated family of face stimuli that vary in distinctiveness. The experiments show that while the "view-mapping" process operates more accurately for typical versus distinctive faces, the base level distinctiveness of the faces is preserved in the face space coding. These data provide insight into how the psychophysically well-established inverse relationship between the typically and recognizability of faces might operate for recognition across changes in viewpoint.

A scheme for learning to grasp objects using visual information is presented. The learning problem is divided into two separate subproblems: choosing grasping points and predicting the quality of a given grasp. For each grasp we store location parameters that code the locations of the grasping points, quality parameters that are relevant features for the assessment of grasp quality, and the associated grade. The location parameters, using a special coding which is not object specific, are used to locate grasping points on new target objects. A function from the quality parameters to the grade is learned from examples. Grasp quality for novel situations can be generalized and estimated using the learned function. An experimental setup using an AdeptOne manipulator was developed to test this scheme. The system had demonstrated an ability to grasp a relatively wide variety of objects, and its performance had significantly improved with practice following a small number of trials. The knowledge learned for a set of objects was successfully generalized to new objects.

How does the brain represent visual objects? In simple perceptual generalization tasks, the human visual system performs as if it represents the stimuli in a low-dimensional metric psychological space (Shepard 1987). In theories of three-dimensional (3D) shape recognition, the role of feature-space representations [as opposed to structural (Biederman 1987) or pictorial (Ullman 1989) descriptions] has long been a major point of contention. If shapes are indeed represented as points in a feature space, patterns of perceived similarity among different objects must reflect the structure of this space. The feature space hypothesis can then be tested by presenting subjects with complex parameterized 3D shapes, and by relating the similarities among subjective representations, as revealed in the response data by multidimensional scaling (Shepard 1980), to the objective parameterization of the stimuli. The results of four such tests, accompanied by computational simulations, support the notion that discrimination among 3D objects may rely on a low-dimensional feature space representation, and suggest that this space may be spanned by explicitly encoded class prototypes.

Nonmetric multidimensional scaling (MDS) is a family of algorithms that allow one to derive a quantitative representation of data from a set of qualitative measurements which must satisfy certain simple constraints. As a tool for vision, MDS combines the advantages of both qualitative and classical approaches, by relying, on the one hand, on an ordinal-scale input representation, and by supporting, on the other hand, the extraction of metric information. The present paper illustrates an application of MDS to the recovery of depth from the rank order of binocular disparity differences for a set of points. Our results indicate that multidimensional scaling constitutes a promising approach to the integration of biological and computational insights into the problem of depth perception.

The maximization of diversity of neuronal response properties has been recently suggested as an organizing principle for the formation of such prominent features of the functional architecture of the brain as the corti(cid:173) cal columns and the associated patchy projection patterns (Malach, 1994). We show that (1) maximal diversity is attained when the ratio of dendritic and axonal arbor sizes is equal to one, as found in many cortical areas and across species (Lund et al., 1993; Malach, 1994), and (2) that maxi(cid:173) mization of diversity leads to better performance in systems of receptive fields implementing steerable/shiftable filters, and in matching spatially distributed signals, a problem that arises in many high-level visual tasks. 1 Anatomical substrate for sampling diversity A fundamental feature of cortical architecture is its columnar organization, mani(cid:173) fested in the tendency of neurons with similar properties to be organized in columns that run perpendicular to the cortical surface. This organization of the cortex was ini(cid:173) tially discovered by physiological experiments (Mouncastle, 1957; Hubel and Wiesel, 1962), and subsequently confirmed with the demonstration of histologically defined columns. Tracing experiments have shown that axonal projections throughout the cerebral cortex tend to be organized in vertically aligned clusters or patches. In par(cid:173) ticular, intrinsic horizontal connections linking neighboring cortical sites, which may extend up to 2 - 3 mm, have a striking tendency to arborize selectively in preferred sites, forming distinct axonal patches 200 - 300 J.lm in diameter. Recently, it has been observed that the size of these patches matches closely the average diameter of individual dendritic arbors of upper-layer pyramidal cells 118 Kalanit Grill Spector, Shimon Edelman, Rafael Malach

Performance of human subjects in a wide variety of early visual processing tasks improves with practice. HyperBF networks (Poggio and Girosi 1990) constitute a mathematically well-founded framework for understanding such improvement in performance, or perceptual learning, in the class of tasks known as visual hyperacuity. The present article concentrates on two issues raised by the recent psychophysical and computational findings reported in Poggio et al. (1992b) and Fahle and Edelman (1992). First, we develop a biologically plausible extension of the HyperBF model that takes into account basic features of the functional architecture of early vision. Second, we explore various learning modes that can coexist within the HyperBF framework and focus on two unsupervised learning rules that may be involved in hyperacuity learning. Finally, we report results of psychophysical experiments that are consistent with the hypothesis that activity-dependent presynaptic amplification may be involved in perceptual learning in hyperacuity.

Previous results on nonlearnability of visual concepts relied on the assumption that such concepts are represented as sets of pixels. The author uses an approach developed by Haussler (1989) to show that under an alternative, feature-based representation, recognition is probably approximately correct (PAC) learnable from a feasible number of examples in a distribution-free manner.>

We describe experiments with a versatile pictorial prototype-based learning scheme for 3-D object recognition. The Generalized Radial Basis Function (GRBF) scheme seems to be amenable to realization in biophysical hardware because the only kind of computation it involves can be effectively carried out by combining receptive fields. Furthermore, the scheme is computationally attractive because it brings together the old notion of a "grandmother" cell and the rigorous approximation methods of regularization and splines.