A Critical Review of Research Related to Object Recognition, Term Paper Example

Introduction

The purpose of this essay is to explore research regarding object recognition in the field of psychology. As measurement and understanding of this concept has evolved considerably over time, this essay aims to provide a critical review of object recognition research. Background information will first be presented, including a basic definition of object recognition. A historical review of research related to object recognition will then be discussed. Particular emphasis will be placed on the neurobiological mechanisms of coding and retrieval, as modern research appears to be trending in this direction. This essay concludes with a brief summary and directions for future research.

Background

This section provides historical background information regarding object recognition. An evolution of research related to the concept is presented, ending with the neurobiological underpinnings of object recognition. This section concludes with a brief summary.

Definition

Object recognition essentially refers to the ability to recognize an object through the utilization of one’s physical senses (Sternberg, 2006). From a cognitive neuroscience perspective, object recognition involves the detection and perception of an object’s physical characteristics, coupled with the attribution of the object’s meaning and purpose (Sternberg, 2006). Object recognition is currently believed to involve a series of progressive steps in which basic physical properties are first detected, these properties are categorized according to familiarity, the visual pattern is combined with information from previous memory, and attributes are formulated based on the object’s meaning and purpose.

The study of object recognition has yielded a wide body of research in the field of psychology and has resulted in the division of theoretical views on how this phenomenon occurs, the physical processes that are responsible for the processing of object information, and the neurobiological mechanisms that underpin the process of coding and retrieval. The following section provides a historical review of research related to object recognition, with a description of major theoretical views.

Research

The study of object recognition has its roots in the mid 1960s, as cognitive perspectives in psychology gained popularity (Sternberg, 2006). The first object recognition studies focused on alignment and geometric primitives (Smith, 2009). Alignment studies sought to understand the object recognition process by aligning images with paired models and detecting changes in recognition systems (Smith, 2009). A geometric primitive utilize basic recognizable shapes used to detect changes in visual orientation. These approaches predominated literature on object recognition until the 1980s. For example, a seminal study utilizing an alignment approach was conducted by a group of researchers at the Massachusetts Institute of Technology devised an alignment experiment in which they paired an object withy an image to detect the process of structuring and organizing information (Huttenlocher & Ullman, 1987). According to these researchers, this approach yields distinct feature labels and avoids structuring recognition as an exponential search process. These early research designs were important in pioneering the concept of artificial intelligence and robotics.

According to the 3-D model representation theory (Marr and Nishihara, 1978), object recognition occurs through a process of pairing representations of the object with a visual model contained in memory. This theory is somewhat similar to the concept of recognition by components (Biederman, 1987), in which individuals are believed to divide objects into their basic geometric components, and then pairing these components with information stored in memory.

These early approaches to object recognition were replaced by viewpoint-based theories in the 1990s (Sternberg, 2006). Examples include invariants, appearance-based methods, the sliding window approach and feature-based theories. According to these theories, object recognition is dependent on the perspective in which it is viewed (Smith, 2009). In the event that a new stimulus is observed, recognition accuracy and timing is delayed. These models were more holistic in their views, and were somewhat contrary to the idea that objects are broken down into their basic geometric elements. According to viewpoint-based theories, objects are stored in memory according to the unique viewpoint in which they are recognized. These approaches present a memory storage problem, however, as an impossible amount of memory would be needed to store the seemingly infinite number of perspectives in which objects are viewed (Smith, 2009).

Some of the memory and shape problems of the representation and viewpoint models were alleviated by the adoption of the combined approach of object recognition in the mid 1990s (Caramazza & Mahon, 2006). According to this theory, object recognition exists on a continuum in which individual viewpoints are available and utilized for varying types of recognition needs. For example, specific recognition mechanisms exist for discriminating between objects, as well as categorizing them (Caramazza & Mahon, 2006).

Present models of object recognition still tend to favor a combination of these localized and global methods of object recognition (Caramazza & Mahon, 2006). Based on these combination approaches, researchers now widely agree that object recognition depends largely on first detecting edges of an object, as well as detecting and processing an objects concavity. Furthermore, nonaccidental properties of the object are important for combining all information together into a recognized form, referred to as a geon (Caramazza & Mahon, 2006). Researchers understand that most information processed is redundant, and that only parts of an object need to be recognized to sense colinearity and identify a geon. Furthermore, only some edges and vertices need to be observed to detect a geon, and only some geons are required to recognize a full shape. Therefore, the specific orientation of the object is relatively unimportant to the recognition process, nor is additional information (Caramazza & Mahon, 2006).

Neurobiological Mechanisms

As the process of object recognition has become increasingly understood, and theoretical approaches to this process have converged, research is trending toward understanding the neural substrates involved in the object recognition process (Sternberg, 2006). Specifically, research is currently exploring the neurobiological mechanisms underlying the process of coding and retrieval (Sternberg, 2006). Research in this area has coincided with advances in cognitive neuroscience and the identification of specific regions of the brain involved in memory, sensory stimulation, and information processing. This line of research has yielded an abundance of sub-theories and perspectives regarding the neurobiological mechanisms related to recognition.

According to Thomson-Schill (2003), the visual component of object recognition can be traced to the dorsal and ventral streams in the brain. The dorsal stream extends through the visual motor cortex, and is believed to play a central role on recognizing objects in space. The dorsal stream functions to provide a visual map for guiding actions. Visual information is processed in the dorsal stream via the occipital lobe, and then this information is transferred into spatial awareness through the parietal lobe (Thomson-Schill, 2003). The ventral stream also plays a key role in object recognition, and is responsible for attaching meaning-information to the object. These pathways were discovered by researchers who studied the motor processing skills of individuals with lesions in different parts of the brain. Researchers noticed that lesions to certain areas disrupted these processing pathways (Thomson-Schill, 2003).

Efforts to understand the neural components of object recognition have placed greater emphasis on the ventral stream (Pavlova et al., 2003). Researchers believe that various aspects of the ventral stream activate in response to object recognition tasks, particularly in the recognition of human faces (Pavlova et al., 2003). For example, the ventral stream is more highly activated in response to a human face than a generic geometric shape (Pavlova et al., 2003). Furthermore, moving objects produce more ventral activation than static objects, as well as in tasks in which the subject is required to unscramble a series of shapes (Pavlova et al., 2003).

A region of the brain known as the lateral occipital complex has been demonstrated to play an important role in recognizing objects, as well as the formulation of perceptions of the object (Thomson-Schill, 2003). Utilizing functional imaging technology, researchers have discovered that familiarity of an object produces activation of the lateral occipital complex, suggesting that this region is important for facilitating the perception and identification process. One theory related to the activation of the lateral occipital complex is that this system works in a top-down fashion, in which certain shapes are selected and produce greater activation in external regions and anterior regions, respectively. This hierarchical approach to object recognition has been replicated in other research (Thomson-Schill, 2003).

Although research has indicated the significance of the lateral occipital complex in recognizing objects, little information regarding semantic attribution has been identified through these studies (Sternberg, 2006). Semantics refers to the forming of relations between objects and attaching meaning and purpose to the stimuli. In studies of individuals with neurological impairments, researchers have discovered pathways that may be involved in the semantic attribution of objects. For example, some researchers have used brain imaging in individuals with Alzheimer’s disease to identify regions of the brain associated with structure and color information. Furthermore, research (e.g., Pavlova et al., 2003) has identified regions of the brain associated with decision-making tasks and the storage of perceptual memories.

One specific aspect of object recognition that influences the degree to which objects can be perceived is the contextual cues associated with the task. The context has been demonstrated to play a central role in both the visual identification of the object, as well as the semantic processing (Pavlova et al., 2003). Research has demonstrated that the degree to which semantic processing regions of the brain are activated in functional imagery studies largely depends on contextual cues, such as the background scene and spatial context (Thomson-Schill, 2003).

Finally, modern research related to object recognition is investigating the neural underpinnings of familiarity and recollection of visual cues. Research has demonstrated that impairments in the ventral stream influence the ability to recognize and memorize an object (Sternberg, 2006). The concept of familiarity can influence semantic attribution, as well as recalling memories. The context of the memory, as well as the viewpoint can influence the speed and accuracy at which objects are recognized or retrieved, and research has pinpointed the ventro-lateral component of the frontal lobe in facilitating this process (Thomson-Schill, 2003). Recalling an object, however, is entirely context-specific, and differs from object familiarity (Thomson-Schill, 2003).

Summary

Object recognition essentially refers to the ability to recognize an object through the utilization of one’s physical senses. The study of object recognition has yielded a wide body of research in the field of psychology and has resulted in the division of theoretical views on how this phenomenon occurs, the physical processes that are responsible for the processing of object information, as and the neurobiological mechanisms that underpin the process of coding and retrieval. Early approaches to object recognition were replaced by viewpoint-based theories in the 1990s. As the process of object recognition has become increasingly understood, and theoretical approaches to this process have converged, research is trending toward understanding the neural substrates involved in the object recognition process.

Conclusions

From a cognitive neuroscience perspective, object recognition involves the detection and perception of an object’s physical characteristics, coupled with the attribution of the object’s meaning and purpose. Object recognition is currently believed to involve a series of progressive steps in which basic physical properties are first detected, these properties are categorized according to familiarity, the visual pattern is combined with information from previous memory, and attributes are formulated based on the object’s meaning and purpose. The study of object recognition has its roots in the mid 1960s, as cognitive perspectives of psychology gained popularity. The first object recognition studies focused on studies of alignment and geometric primitives. These early approaches to object recognition were replaced by viewpoint-based theories in the 1990s. According to these theories, object recognition is dependent on the perspective in which it is viewed.

Present models of object recognition still tend to favor a combination of these localized and global methods of object recognition. Based on these combination approaches, researchers now widely agree that object recognition depends largely on first detecting edges of an object, as well as detecting and processing an objects concavity. Research is currently exploring the neurobiological mechanisms underlying the process of coding and retrieval. Research in this area has coincided with advances in cognitive neuroscience and the identification of specific regions of the brain involved in memory, sensory stimulation, and information processing. The visual component of object recognition can be traced to the dorsal and ventral streams in the brain. The ventral stream plays a key role in object recognition, and is responsible for attaching meaning-information to the object. These pathways were discovered by researchers who studied the motor processing skills of individuals with lesions in different parts of the brain.

A region of the brain known as the lateral occipital complex has been demonstrated to play an important role in recognizing objects, as well as the formulation of perceptions of the object. Although research has indicated the significance of the lateral occipital complex in recognizing object, little information regarding semantic attribution has been identified through these studies. One specific aspect of object recognition that influences the degree to which objects can be perceived is the contextual cues associated with the task. Finally, modern research related to object recognition is investigating the neural underpinnings of familiarity and recollection of visual cues.

The opportunities to advance research in the study of object recognition are wide-ranging. As researchers understand more about the activation and roles of localized regions of the brain involved in recognition, coding, and retrieval, more will be understood about memory storage, recollection, and spatial awareness. Furthermore, the understanding of object recognition may have implications for treating memory deficits such as Alzheimer’s disease, or treating patients with brain injuries and lesions. Finally, efforts to converge varying methodological approaches to the understanding of object recognition will produce more holistic and organized frameworks to fuel future research.

Reference:

Biederman, I. (1987). Recognition by components: A theory of human image understanding. Psychological Review, 94, 115-147.

Caramazza, A., & Mahon, B. Z. (2006). The organization of conceptual knowledge in the brain: the future’s past and some future directions. Cognitive Neuropsychology, 23:1, 13-38.

Huttenlocher, D. P., & Ullman, S. (1987). Object recognition using alignment. Report prepared for the Artificial Intelligence Laboratory of the Massachusetts Institute of Technology.

Marr, D., & Nishihara, H. (1978). Representation and recognition of the spatial organization of three-dimensional shapes. Proceedings of the Royal Society of London , 200, 269-294.

Pavlova, M., Staudt, M., Sokolov, A., Birbaumer, N., & Krageloh-Mann, I. (2003). Perception and production of biological movement in patient with early periventricular brain lesions. Brain, 126, 692 – 701.

Smith, L. B. (2009). From fragments to geometric shape changes in visual object recognitiom between 18 and 24 months. Current Directions in Psychological Science, 18:5, 290-294.

Sternberg, R. J. (2006). Cognitive Psychology (4^th Ed.). Belmont, CA: Thomson Wadsworth.

Thompson-Schill, S. L. (2003). Neuroimaging studies of semantic memory: Inferring “how” from “where”. Neuropsychologia, 41, 280–292.