Research
Semantic space consists of representations, such as objects and concepts, and the relations between them. In the same way that the movement of a ball can be predicted by properties of the landscape, we seek to predict the movement of the "stream of thought" based on local and temporary properties of the semantic space. Moreover, we believe that based on local properties of the semantic space at a given time-point, we can predict how the space will evolve locally over time. Accordingly, we look at the semantic space-time in a "geographical" manner and adopt the notion of distance between concepts in the semantic space. This view allows us to define psychological terms such as level of abstraction and uniqueness geometrically and to describe their connection to the direction of the stream of thought. For our experimental work, we developed a network-oriented tool kit: we built a network and developed exact measures for a series of properties, such as abstraction level, uniqueness, and attractiveness.
We aim to formulate rules governing some aspects of semantics, and believe these rules should be well defined, clear, and definite, like physical laws. In attempting to achieve this goal, we distinguish between three concepts:
(a) Structure: defines the basic properties of the semantic space, which consists only of concepts to which the agent pays attention at a certain moment.
​
(b) Dynamics: defines how the semantic space changes from moment to moment. Our central goal is to characterize and predict changes in momentary structure.
​
(c) Movement– defines the laws that determine the transition of thought from one concept to another (in an ever-changing space). Our main goal is to predict the stream of thought and the dynamics of the space, based on the properties of its structure.
​
The second line of research focuses on episodic memory. Memory research is, and has been, dominated by the study of episodic memory — the encoding and retrieval of recently experienced information. Significantly less attention has been given to semantic memory — our existing knowledge networks. Work in my lab lies at an interface between these two approaches: (I.) How does knowledge activated (implicitly or explicitly) during an episode affect memory of this episode?
To examine this question, we consider (II.) how our existing knowledge (parts of which become activated during an
episode) is organized and stored.
For hearing Professor Maril's interview about creating and recalling memories(in Hebrew), click here
​
​
​