A probabilistic model (see SE 013 578) describing information processing during the cognitive tasks of recall and problem solving was tested, refined, and developed by testing graduate students on a number of tasks which combined oral, written, and overt "input" and "output" modes in several ways. In a verbal chain one subject repeated the substance of a communication to a second person, who in turn relayed the information. Each subject then wrote a statement of what was heard. Other tasks included verbal problem solving and immediate and delayed recall. Between one and one-and-one-half bits of long term memory information were processed in problem solving and recall tasks. Some subjects restructured the terms of a communication during recall and differed in problem solving strategies. There were interactions between information input and information retrieved from long-term memory. Subjects could tolerate 47.6 percent increase of information input before "noise" interfered with problem solving. In recall tasks there was a 59.7 percent decrease of information input. An individual's cognitive processing style can be identified from parameters of the model. (A glossary of terms, definitions of model parameters, and 14 data tables are appended.) (Author/AL)